Framer vs Sketch

Agent ops

1. ai-native userAn agent can create or modify designs programmatically — through a write-capable API, plugin surface, or MCP tools — not just read them

   weight 2 · round drawn
   Framer✓fullclaimed8/10

   Framer documents a write-capable external agent connection (Claude Code, Cursor, Codex) that can create/update/delete CMS items and modify canvas/components, backed by branching for safe review, plus a plugin surface and Server API for programmatic site automation. Missing for 10: independent/hands-on verification of the external-agent write capability and no public OpenAPI/formal API schema was found by the probe.

   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, and decide when to merge …”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items. You stay in control of …”
   • [claimed-docs] “Framer has branching, so every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, …”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Plugins Build small apps that can interact with the Framer Editor and CMS.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project from the tools you already us…”

   proof ↗⚑ flag

   Sketch✓fullclaimed8/10

   Sketch documents multiple write-capable programmatic surfaces: an official JavaScript Plugin API for creating/modifying designs, sketchtool CLI, sketch:// URL scheme for workflow integration, and explicit MCP support letting any MCP-compatible AI client connect to Sketch, including AI features that build out whole frames/pages from a repo. This goes beyond read-only inspection (which is the separate developer-handoff feature) into actual design creation/modification via agent tooling. Missing for 10: independent hands-on verification that MCP tools can actually write/modify designs (evidence is vendor-only), and no detailed spec of exposed MCP tool actions.

   • [claimed-docs] “Write a Sketch plugin in JavaScript. Take advantage of ES6, access macOS frameworks and use the Sketch APIs without learning Objective-C or …”
   • [claimed-docs] “Try the Sketch JavaScript API now, directly within Sketch.”
   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”
   • [claimed-docs] “Take any frame in Sketch and build it out in code with 1:1 visual fidelity.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”

   proof ↗⚑ flag

2. ai-native userAn agent can render and export designs — screenshots, images, or production formats — through a documented programmatic surface

   weight 2 · round to Sketch
   Framer—none0/10

   Evidence covers AI agents editing designs, CMS, and publishing via a Server API and external-agent connections, but nothing documents a programmatic render/export surface for screenshots, images, or production-format exports (no export/render API endpoints, and probes for OpenAPI specs all 404).

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Connect your product and automate publishing with the Framer API.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”

   proof ↗⚑ flag

   Sketch✓fullprobed7/10

   Sketch documents `sketchtool`, a command-line tool explicitly for creating/exporting assets programmatically, plus an official JavaScript plugin API and 'integrate into workflows and server applications' language, giving agents a documented surface to render/export designs; MCP support further lets AI clients drive Sketch. Missing for 10: independent/hands-on verification of sketchtool's export formats and no explicit mention of headless/server rendering outside macOS, and no OpenAPI/REST surface was found (probe 404s).

   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.sketch.com/openapi.json, https://www.sketch.com/swagger.json, https://www.sketch.com/api…”

   proof ↗⚑ flag

3. ai-native userMy coding agent can pull real design context — frames, variables, screenshots, component mappings — through an MCP server or API to implement the design

   weight 3 · round to Sketch
   Framer~partialclaimed7/10

   Framer explicitly markets a native 'External Agent' connection (explicitly positioned as replacing a separate MCP server) that gives coding agents like Claude Code, Cursor, and Codex direct access to canvas, components, CMS, and project context, including read/write CMS operations and branch-based change tracking. This closely matches the story's intent of pulling design context to implement changes, but the specific mention of 'screenshots' and 'variables' from the story isn't explicitly evidenced, and all evidence is first-party docs with no independent/hands-on corroboration of the external agent workflow. Missing for 10: explicit support for screenshots/variables extraction, independent developer confirmation of the external-agent integration working as described.

   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project from the tools you already us…”
   • [claimed-docs] “Connect Claude Code, Codex, or Cursor to update a Framer project.”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items. You stay in control of …”
   • [claimed-docs] “every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, and decide when to merge …”
   • [claimed-docs] “Framer has branching, so every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, …”

   proof ↗⚑ flag

   Sketch✓fullclaimed7/10

   Sketch's own docs explicitly describe an MCP server integration ('Connect any MCP-compatible AI client you like to Sketch') that lets an AI client pull frames and recreate text styles, colors, UI components and pages, then build them out in code with 1:1 visual fidelity — directly matching the story's design-context-to-implementation workflow. Missing for 10: independent/hands-on verification of the MCP server actually working, explicit mention of pulling 'variables' or raw screenshots through the API, and any third-party corroboration beyond vendor marketing pages.

   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”
   • [claimed-docs] “Take any frame in Sketch and build it out in code with 1:1 visual fidelity.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”

   proof ↗⚑ flag

Agent access

1. ai-native userPoint an agent at llms.txt or agent-oriented docs

   weight 2 · round drawn
   Framer✓fullprobed8/10

   A direct probe confirms Framer serves a live llms.txt file (HTTP 200) with structured content describing the product and linking to migration/agent docs, and Framer's own docs explicitly reference this file's content, giving concrete first-party evidence an AI agent can be pointed at it. missing for 10: independent/third-party confirmation that agents successfully consume this llms.txt in practice, and broader agent-oriented doc coverage beyond this single file.

   • [probe] “PROBE llms.txt: HTTP 200 at https://www.framer.com/llms.txt # Framer > Framer is an AI website builder for designers and teams. Generate ed…”
   • [claimed-docs] “Switch to Framer](https://www.framer.com/switch/?md): Migration from another builder.”
   • [claimed-docs] “Framer allows you to host static files directly on your site or custom domain. This feature is available on the **Pro**, **Enterprise**, and…”

   proof ↗⚑ flag

   Sketch✓fullprobed8/10

   Sketch actually publishes an llms.txt file that returns HTTP 200 and serves a structured docs index for agent consumption (getting-started, prototyping, developer-handoff, workspace docs), confirmed by both docs listing and a runtime probe. Missing for 10: no evidence of broader agent-oriented markdown mirrors (the /docs/.md probe 404s) or an OpenAPI/machine-readable schema beyond llms.txt.

   • [claimed-docs] “Learn how Single Sign-On (SSO) and SCIM provisioning work in Sketch, and find step-by-step guides to get set up.”
   • [probe] “PROBE llms.txt: HTTP 200 at https://www.sketch.com/llms.txt # Sketch ## Docs - [Getting started](https://www.sketch.com/docs/getting-start…”
   • [probe] “PROBE runtime (recorded 2026-09-07): https://www.sketch.com/llms.txt served a full docs index ('# Sketch') for agent consumption, covering g…”
   • [probe] “PROBE docs-md: HTTP 404 at https://www.sketch.com/docs/.md”

   proof ↗⚑ flag

2. ai-native userRun the product headlessly / in CI for automation

   weight 2 · round to Sketch
   Framer~partialprobed3/10

   Framer advertises a Server API to 'automate publishing' and integrate with external products/agents, which implies some programmatic/headless control, but there is no documented CLI, headless mode, or CI-specific workflow, and probes show no OpenAPI/API spec is publicly discoverable (404s). Missing for 10: concrete CI/headless setup docs, API reference/spec, and evidence of automated pipeline usage beyond marketing blurbs.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”

   proof ↗⚑ flag

   Sketch~partialcommunity5/10

   Sketch documents a command-line tool (`sketchtool`) for scripted asset creation and claims it can integrate 'into your workflows and server applications, on any platform,' which points at CI/headless automation use. However, this is only first-party doc language with no independent verification, and community evidence confirms Sketch itself is strictly Mac-native and the vendor has refused Windows/Linux support, undercutting the 'any platform' claim. Missing for 10: independent CI/automation examples or case studies, clarity on macOS-only runtime requirements for sketchtool, and confirmation of non-interactive/headless execution in real pipelines.

   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [community] “It's not just their commitment to Desktop, but their commitment specifically to not move off MacOS where they have stated that they 'will no…”

   proof ↗⚑ flag

3. ai-native userPlug MCP servers into this product so it can use their tools

   weight 3 · round drawn
   Framer—none0/10

   Framer explicitly states it does not use or require MCP: 'Framer doesn't require a separate MCP server... without manual MCP setup' (framer-docs-34). Its actual capability runs in the opposite direction — external agents like Claude Code/Cursor connect INTO Framer to use Framer's canvas/CMS as their tool, not Framer consuming external MCP servers' tools. No evidence exists of Framer acting as an MCP client that plugs in third-party MCP servers.

   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [claimed-docs] “Connect Claude Code, Codex, or Cursor to update a Framer project.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project from the tools you already us…”

   proof ↗⚑ flag

   Sketch—none0/10

   Evidence only shows Sketch acting as an MCP *server* that external AI clients (e.g. Claude, Cursor) can connect to and use Sketch's own tools (sketch-docs-18, sketch-docs-34) — the reverse of the story, which asks whether Sketch itself can plug in external MCP servers to consume their tools. No documentation or community evidence indicates Sketch can act as an MCP client consuming third-party tool servers.

   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”

   proof ↗⚑ flag

4. ai-native userConnect an agent via an official MCP server

   weight 3 · round to Sketch
   Framer—none0/10

   Framer explicitly states it does not offer an MCP server, instead providing a proprietary 'External Agent connection' as an alternative to MCP (framer-docs-34). Since the story specifically asks for an official MCP server, and Framer's own docs assert this is unnecessary and unsupported by design, there is no evidence of an actual MCP server implementation.

   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”

   proof ↗⚑ flag

   Sketch✓fullclaimed7/10

   Sketch documents that users can connect any MCP-compatible AI client to Sketch itself, positioning Sketch as an MCP server that AI agents can plug into, with no extra tiers or tokens required. Missing for 10: independent/hands-on verification that the MCP connection works reliably, and detailed technical docs on the server's capabilities/setup beyond marketing copy.

   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”
   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”

   proof ↗⚑ flag

5. ai-native userUse an official CLI

   weight 2 · round to Sketch
   Framer—none0/10

   Evidence shows Framer offers a Server API, plugins, and 'External Agent' connections for tools like Claude Code/Cursor/Codex, but nowhere does it mention an official Framer CLI tool for AI-native workflows.

   ⚑ flag

   Sketch~partialclaimed6/10

   Sketch ships an official CLI, `sketchtool`, documented for asset creation and integration into workflows/server applications on any platform, which agentic/AI tools could invoke. However, there's no evidence this CLI is positioned or extended for AI-native agent workflows specifically (no mention of AI agents using sketchtool, no JSON/agent-friendly output specs, and MCP support is a separate non-CLI integration). Missing for 10: agent-oriented CLI documentation, examples of AI/agent usage of sketchtool, independent hands-on confirmation of its use in automated pipelines.

   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “Integrate Sketch into your workflows with the `sketch://` URL scheme.”
   • [claimed-docs] “Use the Sketch Plugin Manager `skpm` to publish a new plugin or an update of an existing one.”

   proof ↗⚑ flag

6. ai-native userDrive the product through a documented public API

   weight 3 · round to Sketch
   Framer~partialprobed6/10

   Framer documents a Server API for automation/publishing and CMS API access, and explicitly supports external AI agents (Claude Code, Cursor, Codex) reading/writing CMS collections via a documented connection flow (framer-docs-5,17,30,39,40,18,24,34). However, probes could not locate any formal API reference/spec (openapi.json/swagger 404s) or detailed endpoint documentation, and there's no independent/community corroboration of developers actually using the Server API. Missing for 10: a discoverable formal API reference/spec, authentication/rate-limit details, and independent developer testimonials confirming API usability.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project from the tools you already us…”
   • [claimed-docs] “Connect Claude Code, Codex, or Cursor to update a Framer project.”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”

   proof ↗⚑ flag

   Sketch✓fullprobed7/10

   Sketch documents an official vanilla JavaScript plugin API 'fully supported between releases', a CLI (`sketchtool`), a `sketch://` URL scheme for workflow/server integration, and explicit MCP support so any MCP-compatible AI client can connect and drive Sketch — directly matching the AI-native/agentic story. However there's no REST/HTTP API or OpenAPI spec (probe shows 404s for openapi/swagger endpoints), and the API is tied to the Mac app/plugin ecosystem rather than a cloud-callable service. Missing for 10: a REST/HTTP or OpenAPI-documented API, and independent hands-on corroboration of AI clients actually driving Sketch via MCP.

   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “Integrate Sketch into your workflows with the `sketch://` URL scheme.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.sketch.com/openapi.json, https://www.sketch.com/swagger.json, https://www.sketch.com/api…”

   proof ↗⚑ flag

7. ai-native userIssue scoped/least-privilege API credentials for an agent

   weight 2 · round drawn
   Framer—none0/10

   Framer documents a Server API and an 'External Agent' connection that lets tools like Claude Code or Cursor read/write CMS data and canvas content, with branching as a safety net for undoing changes — but nothing describes issuing scoped, least-privilege, or permission-limited API credentials/tokens for agents. Missing for 10: any mention of API key scopes, granular permission levels, token expiration/rotation, or role-based credential issuance for agent access.

   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, and decide when to merge …”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items. You stay in control of …”
   • [claimed-docs] “Framer has branching, so every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, …”

   proof ↗⚑ flag

   Sketch—none0/10

   Evidence shows MCP connectivity, SSO, SCIM provisioning, and permissions groups, but nothing about issuing scoped or least-privilege API credentials/tokens specifically for agents; permissions groups and SSO are user-level access controls, not credential scoping mechanisms. Missing for 10: any documentation of API tokens, scoped keys, or credential issuance workflow tailored to agent use.

   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “SCIM provisioning”
   • [claimed-docs] “Permissions groups”
   • [claimed-docs] “Single Sign-On (SSO) * Project archiving * Permissions groups”

   proof ↗⚑ flag

8. ai-native userBuild against official SDKs

   weight 2 · round drawn
   Framer~partialprobed6/10

   Framer documents a Server API, plugin SDK, code components/overrides, and an external agent connection for building integrations against official interfaces, but there's no discoverable OpenAPI spec (probe found 404s across candidate paths) and no independent/hands-on developer corroboration of building against these SDKs. missing for 10: publicly discoverable API reference/OpenAPI spec, independent developer corroboration, SDK versioning/language support details.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Plugins Build small apps that can interact with the Framer Editor and CMS.”
   • [claimed-docs] “Code Overrides Higher Order Components to modify Layer and Component properties.”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”

   proof ↗⚑ flag

   Sketch~partialprobed6/10

   Sketch publishes an official vanilla JavaScript API, a CLI (sketchtool), plugin manager (skpm), and a documented sketch:// URL scheme for integrating into workflows and server apps, which constitute genuine official SDKs/APIs. However, there's no OpenAPI spec, no REST API, and probes confirm no openapi.json exists; the AI-native story is mostly served via MCP connectivity rather than a broad official SDK ecosystem. Missing for 10: a formal REST/OpenAPI API, multi-language SDKs beyond JS, and independent developer corroboration of building production integrations against these SDKs.

   • [claimed-docs] “Write a Sketch plugin in JavaScript. Take advantage of ES6, access macOS frameworks and use the Sketch APIs without learning Objective-C or …”
   • [claimed-docs] “Try the Sketch JavaScript API now, directly within Sketch.”
   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “Integrate Sketch into your workflows with the `sketch://` URL scheme.”
   • [claimed-docs] “Use the Sketch Plugin Manager `skpm` to publish a new plugin or an update of an existing one.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.sketch.com/openapi.json, https://www.sketch.com/swagger.json, https://www.sketch.com/api…”

   proof ↗⚑ flag

9. ai-native userSubscribe to events via webhooks

   weight 2 · round drawn
   Framer—none0/10

   Evidence covers Framer's Server API, external agent connections, and plugins, but there is no mention of webhooks or event subscription mechanisms anywhere in the docs or probes; OpenAPI probe also returned 404s with no webhook spec found.

   ⚑ flag

   Sketch—none0/10

   No evidence of Sketch offering webhooks or event subscription mechanisms; the product exposes a plugin API, CLI (sketchtool), URL scheme, and MCP integration for AI clients, but nothing about webhook subscriptions for events. This is a fair axis for a platform with an API/integration ecosystem, so absence of evidence yields 'none'.

   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “Integrate Sketch into your workflows with the `sketch://` URL scheme.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”

   proof ↗⚑ flag

Agentic features

1. ai-native userGet AI-generated insights and suggestions from my data inside the product

   weight 2 · round to Sketch
   Framer~partialclaimed4/10

   Framer's AI agent can restructure and suggest layout changes based on generic 'conversion best practices' (framer-docs-31/38) and manages/organizes CMS data (framer-docs-8/46), which is adjacent to data-driven suggestions, but there is no evidence of the agent analyzing a user's actual site/traffic data to surface analytical insights (e.g., performance, SEO, or visitor-behavior insights) — it is generative/design assistance rather than data-insight generation. Missing for 10: evidence of analytics-based insight generation, data pattern analysis, or dashboards summarizing user's own data trends.

   • [claimed-docs] “Reorganize the homepage flow based on conversion best practices. Move social proof higher, put pricing before the FAQ”
   • [claimed-docs] “Reorganize the homepage flow based on conversion best practices. Move social proof higher, put pricing before the FAQ, and make sure every s…”
   • [claimed-docs] “The CMS agent sets up, organizes, and updates your entire CMS, connecting it directly to the canvas so content and design stay in sync.”
   • [claimed-docs] “Manage more. Publish faster. The CMS agent sets up, organizes, and updates your entire CMS, connecting it directly to the canvas so content …”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”

   proof ↗⚑ flag

   Sketch~partialclaimed6/10

   Sketch's AI features (docs-14,15,26,27,39) generate suggestions like color palettes, dark-mode variants, layer organization, and placeholder content directly from design data, which counts as AI-generated insights/suggestions inside the product; MCP integration (docs-18,34) also lets AI clients query/act on Sketch data. However, these are largely generative/automation utilities rather than analytic 'insights' about the user's data, and there is no independent/hands-on corroboration of quality or accuracy. missing for 10: independent verification of AI feature accuracy/usefulness, broader insight-generation (e.g., analytics on design usage/trends) beyond visual generation tasks, and clearer separation of AI-insight vs AI-automation claims.

   • [claimed-docs] “Switch light mode components for their dark mode counterparts — all without making a single manual swap.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it. An instant color palette, ready to use in your document.”
   • [claimed-docs] “Sort, rename, and align layers automatically. Wrangle icon sets, keep Libraries clean, and bring calm to chaotic files.”
   • [claimed-docs] “Replace placeholder text across every layer with true-to-life, context-aware content. No more lorem ipsum.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”

   proof ↗⚑ flag

2. ai-native userSet up automations that run autonomously in the background

   weight 2 · round to Framer
   Framer~partialclaimed4/10

   Framer offers a Server API to 'automate publishing' and lets external agents (Claude Code, Cursor, Codex) connect to read/write CMS and canvas content, which could be wired into autonomous workflows, but all documented agent interactions (framer-docs-6/7/26) are conversational/on-demand rather than scheduled or self-running background automations. There's no evidence of triggers, schedules, or persistent background jobs — only API/connection primitives a developer would need to build such automation themselves. missing for 10: evidence of scheduled/triggered background jobs, autonomous run history or logs, native workflow/automation builder distinct from on-demand AI agent editing.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project from the tools you already us…”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”

   proof ↗⚑ flag

   Sketch—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

3. ai-native userDelegate tasks to a built-in AI assistant inside the product

   weight 3 · round to Framer
   Framer✓fullclaimed7/10

   Framer ships multiple built-in AI agents (design agent, CMS agent, AI website generator) that users can delegate tasks to directly inside the product via natural-language prompts (e.g. 'make responsive', 'switch to dark mode', CMS setup) as shown in framer-docs-6,7,8,26,42,43,44,45,46. Missing for 10: independent/hands-on user reports specifically validating the built-in AI assistant's reliability and quality (community evidence available only covers older, unrelated Framer features like animations).

   • [claimed-docs] “Framer’s AI agent creates editable pages, sections, copy, and visuals directly in your project.”
   • [claimed-docs] “Describe the look you want. The agent takes care of colors, fonts, component styles, motion and effects across your site.”
   • [claimed-docs] “The CMS agent sets up, organizes, and updates your entire CMS, connecting it directly to the canvas so content and design stay in sync.”
   • [claimed-docs] “Framer’s AI agent creates editable pages, sections, copy, and visuals directly in your project. Keep refining in the conversation, or take o…”
   • [claimed-docs] “Start from a prompt and use the Framer canvas to refine layouts, typography, components, breakpoints, spacing, color, and effects. AI helps …”
   • [claimed-docs] “Ask the agent to “Make this responsive for tablet and phone.” It adapts your layouts across breakpoints, adjusting stacks, type sizes, spaci…”
   • [claimed-docs] “Drop in images or share a URL and the agent uses it as a starting point. The layout, style, and feel of your reference, brought directly ont…”
   • [claimed-docs] “Sticky navbars, animated menus, interactive carousels, just describe what you want. The agent handles the wiring so you can focus on the des…”
   • [claimed-docs] “Manage more. Publish faster. The CMS agent sets up, organizes, and updates your entire CMS, connecting it directly to the canvas so content …”

   proof ↗⚑ flag

   Sketch~partialclaimed4/10

   Sketch.com/ai lists several AI-powered automation features (auto dark-mode conversion, palette generation from a photo, automatic layer renaming/sorting, context-aware placeholder text) that a user can invoke to offload discrete design tasks, which is a form of task delegation. However, these are framed as one-off automated tools rather than a conversational, built-in assistant that takes freeform instructions, and the closest thing to an 'assistant' (docs-16/18/34) actually describes connecting an *external* MCP-compatible AI client to Sketch, not a native in-app assistant. Missing for 10: evidence of a conversational/chat-style built-in assistant, evidence of multi-step task delegation beyond single-purpose generators, and independent hands-on confirmation of these AI features working as described.

   • [claimed-docs] “Switch light mode components for their dark mode counterparts — all without making a single manual swap.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it. An instant color palette, ready to use in your document.”
   • [claimed-docs] “Sort, rename, and align layers automatically. Wrangle icon sets, keep Libraries clean, and bring calm to chaotic files.”
   • [claimed-docs] “Replace placeholder text across every layer with true-to-life, context-aware content. No more lorem ipsum.”
   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”

   proof ↗⚑ flag

4. ai-native userOperate the product with natural-language commands

   weight 2 · round to Framer
   Framer✓fullclaimed8/10

   Framer's docs show extensive natural-language operation: the built-in AI agent creates pages/sections/copy from prompts, restyles sites ('switch to dark mode', 'use Fraunces/Inter'), reorganizes layouts, and makes sites responsive purely from described intent, plus a CMS agent and external-agent integration (Claude Code/Cursor/Codex) that reads/writes CMS and canvas content via natural language, with branch-based safety for changes. missing for 10: independent/hands-on third-party corroboration of the AI/CMS agent actually working as described (community evidence only covers older non-AI Framer features).

   • [claimed-docs] “Framer’s AI agent creates editable pages, sections, copy, and visuals directly in your project.”
   • [claimed-docs] “Describe the look you want. The agent takes care of colors, fonts, component styles, motion and effects across your site.”
   • [claimed-docs] “The CMS agent sets up, organizes, and updates your entire CMS, connecting it directly to the canvas so content and design stay in sync.”
   • [claimed-docs] “Update the entire site’s typography. Use Fraunces for headings and Inter for body text.”
   • [claimed-docs] “Reorganize the features section from a list layout to a 2x3 grid of feature cards.”
   • [claimed-docs] “Switch the site from light mode to dark mode with dark backgrounds, light text, and updated component colors throughout.”
   • [claimed-docs] “Reorganize the homepage flow based on conversion best practices. Move social proof higher, put pricing before the FAQ”
   • [claimed-docs] “Update the entire site to use our new brand colors: primary #1A1A2E, accent #E94560, and light background #F5F5F5.”
   • [claimed-docs] “Ask the agent to “Make this responsive for tablet and phone.” It adapts your layouts across breakpoints, adjusting stacks, type sizes, spaci…”
   • [claimed-docs] “Drop in images or share a URL and the agent uses it as a starting point. The layout, style, and feel of your reference, brought directly ont…”
   • [claimed-docs] “Sticky navbars, animated menus, interactive carousels, just describe what you want. The agent handles the wiring so you can focus on the des…”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, and decide when to merge …”

   proof ↗⚑ flag

   Sketch~partialclaimed6/10

   Sketch documents AI features (auto dark-mode conversion, color palette generation, layer renaming, placeholder text replacement) and explicit MCP-client connectivity so 'any MCP-compatible AI client' can drive design tasks, implying natural-language-driven operation via an external agent. However there is no first-party evidence of an in-app chat/command interface where a user directly types natural-language instructions to Sketch itself — the AI capabilities are either fixed one-click features or routed through external MCP clients. Missing for 10: a native in-app natural-language command interface, hands-on/independent verification that MCP-driven natural-language commands actually work end-to-end.

   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”
   • [claimed-docs] “Switch light mode components for their dark mode counterparts — all without making a single manual swap.”
   • [claimed-docs] “Replace placeholder text across every layer with true-to-life, context-aware content. No more lorem ipsum.”
   • [claimed-docs] “Sort, rename, and align layers automatically. Wrangle icon sets, keep Libraries clean, and bring calm to chaotic files.”

   proof ↗⚑ flag

Api quality

1. ai-native userExplore an interactive API reference with runnable examples

   weight 2 · round to Sketch
   Framer—none0/10

   Framer documents a Server API, Plugins API, and Code Components, but nothing shows an interactive API reference with runnable/try-it examples; probe evidence explicitly found no OpenAPI/swagger spec (404s at all candidate paths), suggesting no such interactive reference exists.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”

   proof ↗⚑ flag

   Sketch~partialprobed3/10

   Sketch documents a JavaScript plugin API and invites developers to 'Try the Sketch JavaScript API now, directly within Sketch,' suggesting some interactive API exploration, but there is no evidence of a full interactive reference with runnable code examples (OpenAPI/swagger probes all 404, no dedicated API-explorer docs found). missing for 10: dedicated interactive API reference UI, runnable code examples/sandbox, OpenAPI/swagger spec, independent confirmation of the 'try now' feature.

   • [claimed-docs] “Try the Sketch JavaScript API now, directly within Sketch.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.sketch.com/openapi.json, https://www.sketch.com/swagger.json, https://www.sketch.com/api…”

   proof ↗⚑ flag

2. ai-native userDownload a machine-readable API spec (OpenAPI or equivalent)

   weight 2 · round drawn
   Framer—none0/10

   Framer offers a Server API and llms.txt, but the probe explicitly found no OpenAPI/swagger spec at any standard location, and no docs mention a downloadable machine-readable API spec.

   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”
   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”

   proof ↗⚑ flag

   Sketch—none0/10

   Direct probing found no OpenAPI/Swagger spec at any standard location (all candidates 404), and no docs reference a machine-readable API spec; Sketch offers a JS plugin API and MCP integration but no downloadable OpenAPI-equivalent.

   • [probe] “PROBE openapi: all candidate paths 404 (https://www.sketch.com/openapi.json, https://www.sketch.com/swagger.json, https://www.sketch.com/api…”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”

   proof ↗⚑ flag

3. ai-native userTest against a sandbox environment without touching production data

   weight 1 · round to Framer
   Framer~partialclaimed6/10

   Framer offers staging environments on a framer.app domain and automatic branching so AI agent changes happen in isolation from the live/main project, which functions as a sandbox against production data (framer-docs-10, framer-docs-11, framer-docs-25, framer-docs-36). However, this is framed around publishing workflow rather than an explicit 'sandbox for AI testing' concept, and there's no independent/hands-on verification of how well this isolation holds up in practice. missing for 10: independent corroboration of branch/staging isolation, explicit documentation framing this as an AI-testing sandbox rather than a publishing safety net, and detail on whether CMS data changes by agents are also fully isolated from production before merge.

   • [claimed-docs] “Staging environments. Test changes securely on a framer.app domain before pushing live.”
   • [claimed-docs] “Branching. Explore ideas in a branch without touching the main project. Merge when it’s ready to ship.”
   • [claimed-docs] “every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, and decide when to merge …”
   • [claimed-docs] “Framer has branching, so every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, …”
   • [claimed-docs] “Branches. Work on a copy of your project and keep the original clean.”

   proof ↗⚑ flag

   Sketch—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

4. ai-native userRely on versioned APIs with a documented deprecation policy

   weight 2 · round drawn
   Framer—none0/10

   No evidence of API versioning scheme or a documented deprecation policy; probes for OpenAPI specs returned 404s and the Server/Fetch API docs mention no versioning or deprecation terms.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Connect your product and automate publishing with the Framer API.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”

   proof ↗⚑ flag

   Sketch—none0/10

   Sketch exposes a JavaScript plugin API, sketchtool CLI, and MCP integration, and claims the JS API is 'fully supported between releases' (sketch-docs-44), but there is no documented API versioning scheme or deprecation policy — probes for OpenAPI specs return 404 and no changelog/deprecation docs are present in the evidence pack.

   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.sketch.com/openapi.json, https://www.sketch.com/swagger.json, https://www.sketch.com/api…”

   proof ↗⚑ flag

Ai

1. designerBuilt-in AI handles asset chores — image generation, background removal, content-aware fills, copy suggestions — inside the editor

   weight 1 · round to Sketch
   Framer~partialclaimed3/10

   Framer's AI agent can generate pages, sections, copy, and visuals (framer-docs-6, framer-docs-26, framer-docs-42) and adjust layout/typography, which touches copy suggestions and some visual generation, but there is no evidence of dedicated asset-chore tools like image generation, background removal, or content-aware fill inside the editor. missing for 10: dedicated image generation tool, background removal feature, content-aware fill capability, hands-on/independent verification of these specific asset tools.

   • [claimed-docs] “Framer’s AI agent creates editable pages, sections, copy, and visuals directly in your project.”
   • [claimed-docs] “Framer’s AI agent creates editable pages, sections, copy, and visuals directly in your project. Keep refining in the conversation, or take o…”
   • [claimed-docs] “Start from a prompt and use the Framer canvas to refine layouts, typography, components, breakpoints, spacing, color, and effects. AI helps …”
   • [claimed-docs] “Describe the look you want. The agent takes care of colors, fonts, component styles, motion and effects across your site.”

   proof ↗⚑ flag

   Sketch~partialclaimed5/10

   Sketch's AI features (sketch-docs-14,15,26,27,39) cover contextual copy replacement, automatic layer organization, dark-mode conversion, and color-palette extraction from a photo, and it can connect to MCP AI clients for code generation—but there is no evidence of built-in image generation or background removal tools. missing for 10: image generation, background/foreground removal, content-aware fill (beyond text), independent hands-on verification of AI feature quality.

   • [claimed-docs] “Switch light mode components for their dark mode counterparts — all without making a single manual swap.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it. An instant color palette, ready to use in your document.”
   • [claimed-docs] “Sort, rename, and align layers automatically. Wrangle icon sets, keep Libraries clean, and bring calm to chaotic files.”
   • [claimed-docs] “Replace placeholder text across every layer with true-to-life, context-aware content. No more lorem ipsum.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”

   proof ↗⚑ flag

2. designerGenerate or edit designs from a natural-language prompt — first drafts, layouts, or full mockups the tool actually produces

   weight 2 · round to Framer
   Framer✓fullclaimed8/10

   Framer's docs describe a native AI agent that generates editable pages, sections, copy, and visuals from prompts, with examples of full mockup edits (typography, grid layouts, brand colors, dark mode, responsive breakpoints, image-based starting points), and lets designers keep refining in canvas after generation. This directly matches the story of prompt-driven first drafts/layouts/mockups, though evidence is entirely vendor-sourced with no independent hands-on verification of AI-generation quality — missing for 10: independent/hands-on confirmation that generated designs are production-quality, and any documented limitations of the AI design agent.

   • [claimed-docs] “Framer’s AI agent creates editable pages, sections, copy, and visuals directly in your project.”
   • [claimed-docs] “Describe the look you want. The agent takes care of colors, fonts, component styles, motion and effects across your site.”
   • [claimed-docs] “Framer’s AI agent creates editable pages, sections, copy, and visuals directly in your project. Keep refining in the conversation, or take o…”
   • [claimed-docs] “Start from a prompt and use the Framer canvas to refine layouts, typography, components, breakpoints, spacing, color, and effects. AI helps …”
   • [claimed-docs] “Ask the agent to “Make this responsive for tablet and phone.” It adapts your layouts across breakpoints, adjusting stacks, type sizes, spaci…”
   • [claimed-docs] “Drop in images or share a URL and the agent uses it as a starting point. The layout, style, and feel of your reference, brought directly ont…”
   • [claimed-docs] “Sticky navbars, animated menus, interactive carousels, just describe what you want. The agent handles the wiring so you can focus on the des…”
   • [claimed-docs] “Update the entire site’s typography. Use Fraunces for headings and Inter for body text.”
   • [claimed-docs] “Reorganize the features section from a list layout to a 2x3 grid of feature cards.”
   • [claimed-docs] “Switch the site from light mode to dark mode with dark backgrounds, light text, and updated component colors throughout.”
   • [claimed-docs] “Update the entire site to use our new brand colors: primary #1A1A2E, accent #E94560, and light background #F5F5F5.”

   proof ↗⚑ flag

   Sketch~partialclaimed4/10

   Sketch's AI page describes narrow, tool-specific AI features (auto light/dark mode conversion, photo-to-palette, code-to-design 1:1 fidelity, placeholder text replacement, layer auto-organization, and MCP-client integration to 'recreate text styles, colors, UI components, and even entire pages') rather than an open natural-language prompt that a designer types to generate an original first-draft layout or mockup from scratch. This is closer to targeted AI utilities layered onto existing designs than a general prompt-to-mockup generator. Missing for 10: evidence of a text-box/chat interface where a designer types a free-form prompt (e.g. 'design a pricing page') and Sketch outputs an original layout/mockup, and any hands-on or independent verification of AI generation quality.

   • [claimed-docs] “Switch light mode components for their dark mode counterparts — all without making a single manual swap.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it.”
   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”
   • [claimed-docs] “Take any frame in Sketch and build it out in code with 1:1 visual fidelity.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it. An instant color palette, ready to use in your document.”
   • [claimed-docs] “Sort, rename, and align layers automatically. Wrangle icon sets, keep Libraries clean, and bring calm to chaotic files.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”
   • [claimed-docs] “Replace placeholder text across every layer with true-to-life, context-aware content. No more lorem ipsum.”

   proof ↗⚑ flag

1. ai-native userPerform bulk operations across many items at once

   weight 2 · round to Framer
   Framer~partialclaimed6/10

   Framer docs show AI-agent driven bulk operations such as site-wide typography/color/theme changes across all pages (framer-docs-19,27,37), CSV bulk import of CMS collections with row-level linking (framer-docs-9,21), and external agents able to create/update/delete many CMS items via API (framer-docs-18,24,35). This supports bulk operations across many items in CMS/design contexts, but there's no documentation of scale limits, batch API endpoints, or performance at large volumes. Missing for 10: explicit bulk/batch API documentation with scale guarantees, independent verification of bulk operations succeeding at scale, and rate-limit/performance details.

   • [claimed-docs] “Added two linked collections: 1. Articles — 39 CSV rows imported. 2. Categories — 10 rows added and linked to Articles as multi-select tags.”
   • [claimed-docs] “Update the entire site’s typography. Use Fraunces for headings and Inter for body text.”
   • [claimed-docs] “use-cases.csv to create Use Case Articles and Categories collections. Populate fields and match each article to a category.”
   • [claimed-docs] “Switch the site from light mode to dark mode with dark backgrounds, light text, and updated component colors throughout.”
   • [claimed-docs] “Update the entire site to use our new brand colors: primary #1A1A2E, accent #E94560, and light background #F5F5F5.”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items. You stay in control of …”

   proof ↗⚑ flag

   Sketch~partialclaimed5/10

   Sketch's AI features include layer-level bulk actions (auto sort/rename/align layers, replacing placeholder text across every layer, swapping light/dark components) and its CLI/plugin API (`sketchtool`, JS API) allow scripted operations across assets, which supports some automation-at-scale. However, there's no evidence of explicit multi-document/multi-project bulk operations or a documented batch API for large-scale item processing beyond within-document layers. Missing for 10: cross-file/cross-project batch operations, explicit bulk API endpoints, and independent verification of scale performance.

   • [claimed-docs] “Sort, rename, and align layers automatically. Wrangle icon sets, keep Libraries clean, and bring calm to chaotic files.”
   • [claimed-docs] “Replace placeholder text across every layer with true-to-life, context-aware content. No more lorem ipsum.”
   • [claimed-docs] “Switch light mode components for their dark mode counterparts — all without making a single manual swap.”
   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”

   proof ↗⚑ flag

2. ai-native userDefine rules that trigger actions automatically on events

   weight 3 · round drawn
   Framer—none0/10

   Evidence shows Framer's API, plugins, and AI agents that can be manually invoked to modify a site, and an API to 'automate publishing,' but nothing describes a rule engine or event-trigger system where a user defines 'when X happens, do Y' automation. missing for 10: any event-trigger/rule definition mechanism, conditional logic, or workflow automation builder within Framer.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Connect your product and automate publishing with the Framer API.”

   proof ↗⚑ flag

   Sketch—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

3. ai-native userVersion, review, and roll back my automations

   weight 1 · round to Framer
   Framer✓fullclaimed8/10

   Framer documents automatic version history with instant rollback, plus branching that isolates every external-agent change so it can be reviewed, undone, or merged — directly matching version/review/rollback for automations run by AI agents. Missing for 10: independent/hands-on confirmation of rollback and branch-merge workflows in practice, and detail on granular diff/review UI beyond marketing docs.

   • [claimed-docs] “Version history and rollback. Every site change is saved automatically, so you can restore any version in seconds.”
   • [claimed-docs] “Every site change is saved automatically, so you can restore any version in seconds.”
   • [claimed-docs] “every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, and decide when to merge …”
   • [claimed-docs] “Framer has branching, so every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, …”
   • [claimed-docs] “Branching. Explore ideas in a branch without touching the main project. Merge when it’s ready to ship.”
   • [claimed-docs] “Staging environments. Test changes securely on a framer.app domain before pushing live.”

   proof ↗⚑ flag

   Sketch~partialclaimed4/10

   Sketch offers document-level version history via 'starred versions' that lets you track changes, control which versions are visible/published, and effectively roll back to prior states, which could cover AI-generated design changes; but there is no evidence of a dedicated 'automation' concept (e.g., saved AI workflows/scripts) with its own versioning, diff/review, or rollback mechanism distinct from general document history. missing for 10: evidence of versioning/rollback specifically for AI-generated automations/workflows (not just document snapshots), a review/diff UI for AI changes, and any audit trail tied to MCP/AI actions.

   • [claimed-docs] “starring a version will publish any changes as a Library update. Any unstarred versions after that won’t become Library updates.”
   • [claimed-docs] “When you star a version in a Library document, anyone using that Library will receive a Library update.”
   • [claimed-docs] “With starred versions, you can indicate important updates in a document's history, set which versions Viewers and Guests can see”
   • [claimed-docs] “With starred versions, you can indicate important updates in a document’s history, set which versions Viewers and Guests can see, and contro…”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”

   proof ↗⚑ flag

Collaboration

1. product managerComment directly on designs, mention teammates, and resolve threads so review happens in the file, not in email

   weight 2 · round drawn
   Framer—none0/10

   Evidence covers branching, version history, AI agents, CMS, and plugins, but nothing about in-file commenting, @mentions of teammates, or resolving comment threads for design review. Missing for 10: any mention of a commenting feature, mention/notification system, or thread resolution workflow.

   ⚑ flag

   Sketch—none0/10

   The evidence pack covers developer handoff, prototyping, symbols, versioning, plugins, and AI features, but contains no mention of in-file commenting, @mentions, or thread resolution—core to this collaboration story. Community notes even highlight Sketch's historical lag on real-time collaboration features versus competitors, but nothing confirms comment/mention/resolve functionality exists.

   • [community] “Sketch is great software, well built, but the fact that in late 2020 they still have to write 'And soon, the ability to collaborate in real-…”

   proof ↗⚑ flag

2. designerMy whole team can edit the same file simultaneously with live cursors and instant sync

   weight 3 · round drawn
   Framer—none0/10

   The evidence pack includes docs on branching, version history, and staging (framer-docs-11, framer-docs-47) but contains no mention of real-time multiplayer editing, live cursors, or instant sync between simultaneous editors — the actual claim in the story. Missing for 10: any documentation of live cursors, simultaneous multi-user canvas editing, or real-time presence indicators.

   ⚑ flag

   Sketch—none0/10

   No evidence in the pack documents live-cursor or simultaneous real-time multiplayer editing in Sketch; the only related community comment explicitly notes that as of late 2020 Sketch still lacked real-time collaboration ('the ability to collaborate in real-time' was still 'soon'), and Figma is cited as having this as its 'killer USP' that Sketch lacks.

   • [community] “Sketch is great software, well built, but the fact that in late 2020 they still have to write 'And soon, the ability to collaborate in real-…”
   • [community] “A browser-based version of Sketch would be huge - this is Figma's killer USP at the moment.”

   proof ↗⚑ flag

Design systems

1. designerManage design tokens or variables — colors, spacing, typography — with modes or themes applied across designs

   weight 2 · round to Sketch
   Framer—none0/10

   Evidence shows Framer's AI agent can restyle colors/typography/light-dark mode on request (framer-docs-27, framer-docs-37, framer-docs-19), but there is no documentation of an actual design-token or variables system with reusable modes/themes applied consistently across designs, akin to Figma variables. Missing for 10: token/variable definition, mode-based theme switching infrastructure, cross-file/theme propagation evidence.

   • [claimed-docs] “Switch the site from light mode to dark mode with dark backgrounds, light text, and updated component colors throughout.”
   • [claimed-docs] “Update the entire site to use our new brand colors: primary #1A1A2E, accent #E94560, and light background #F5F5F5.”
   • [claimed-docs] “Update the entire site’s typography. Use Fraunces for headings and Inter for body text.”

   proof ↗⚑ flag

   Sketch~partialcommunity5/10

   Sketch supports reusable Styles and Symbols for colors, text, and components across designs, and its AI features can generate named color variables and swap light/dark mode components automatically, giving some token/theme-like behavior. However, there is no evidence of a first-class 'variables with modes' system (e.g., defining semantic tokens with multiple named modes/themes applied project-wide) comparable to dedicated design-token tooling. Missing for 10: explicit variables/tokens feature with mode switching across spacing/typography, first-party documentation of a mode system, and independent hands-on confirmation of theme application at scale.

   • [claimed-docs] “When you make changes to a Symbol, those changes appear everywhere you use it in your designs.”
   • [claimed-docs] “Switch light mode components for their dark mode counterparts — all without making a single manual swap.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it.”
   • [claimed-docs] “Drop in a photo and get a full set of named color variables that complement it. An instant color palette, ready to use in your document.”
   • [claimed-docs] “Symbols can speed up your workflow by giving you a way to save and reuse common elements across your designs.”
   • [claimed-docs] “Symbols can speed up your workflow by giving you a way to save and reuse common elements across your designs. When you make changes to a Sym…”
   • [community] “Sketch is great as a design system manager. It's not all that great as a sketching tool (in spite of the name) and is a terrible photo edito…”

   proof ↗⚑ flag

2. designerBuild reusable components with variants and per-instance overrides that stay linked to their source

   weight 3 · round to Sketch
   Framer~partialclaimed3/10

   Evidence only touches tangentially on component reuse — Code Components with visual controls and Code Overrides (HOCs to modify Layer/Component properties) hint at reusable, overridable components, but nothing explicitly describes variants or per-instance overrides that stay linked to a master component/source as in a design-system workflow. Missing for 10: explicit documentation of component 'variants', proof that per-instance overrides remain linked/syncable to the source component, and any hands-on/community confirmation of this specific workflow.

   • [claimed-docs] “Code Components Custom React Components with visual controls to render anything on your site.”
   • [claimed-docs] “Code Overrides Higher Order Components to modify Layer and Component properties.”

   proof ↗⚑ flag

   Sketch~partialcommunity6/10

   Sketch's Symbols documentation confirms reusable components that stay linked to their source (edits propagate everywhere) and supports nested Symbols for complex structures, and community reviews praise Sketch's DRY, component-based design system workflow. However, the evidence never explicitly documents a 'variants' feature or per-instance overrides (e.g., swappable states/overrides on symbol instances) as Figma-style components do. Missing for 10: explicit documentation of variant properties and per-instance override controls, independent hands-on confirmation of override behavior.

   • [claimed-docs] “When you make changes to a Symbol, those changes appear everywhere you use it in your designs.”
   • [claimed-docs] “Nested Symbols give you finer control over complex designs.”
   • [claimed-docs] “Symbols can speed up your workflow by giving you a way to save and reuse common elements across your designs.”
   • [claimed-docs] “Symbols can speed up your workflow by giving you a way to save and reuse common elements across your designs. When you make changes to a Sym…”
   • [community] “Sketch is great as a design system manager. It's not all that great as a sketching tool (in spite of the name) and is a terrible photo edito…”
   • [community] “It's light, fast, cheap, easy to use and has excellent support for reusable components, making working with design systems a breeze. Adobe h…”

   proof ↗⚑ flag

3. designerShared libraries publish components and styles across files and projects, with controlled updates when the source changes

   weight 2 · round to Sketch
   Framer—none0/10

   Framer's evidence covers code components, plugins, branching, and CMS collections within a single project, but there is no documentation of a shared/team library mechanism that publishes components or styles across separate files or projects with controlled update propagation (the Figma-style library model). Missing for 10: any mention of a cross-project component/style library, library publishing workflow, or update-notification/accept-update mechanism.

   • [claimed-docs] “Code Components Custom React Components with visual controls to render anything on your site.”
   • [claimed-docs] “Code Overrides Higher Order Components to modify Layer and Component properties.”
   • [claimed-docs] “Plugins Build small apps that can interact with the Framer Editor and CMS.”

   proof ↗⚑ flag

   Sketch✓fullcommunity7/10

   Sketch's docs explicitly describe Libraries where starring a version publishes changes as a controlled Library update, letting Editors receive updates while Viewers/Guests see only chosen versions, and Symbols propagate style/component changes across all instances. Community commentary corroborates Sketch's strength as a 'design system manager' built around reusable symbols/libraries. Missing for 10: independent hands-on verification of multi-project library update propagation at scale, and documentation of failure modes/conflicts when libraries diverge.

   • [claimed-docs] “starring a version will publish any changes as a Library update. Any unstarred versions after that won’t become Library updates.”
   • [claimed-docs] “When you star a version in a Library document, anyone using that Library will receive a Library update.”
   • [claimed-docs] “With starred versions, you can indicate important updates in a document's history, set which versions Viewers and Guests can see”
   • [claimed-docs] “With starred versions, you can indicate important updates in a document’s history, set which versions Viewers and Guests can see, and contro…”
   • [claimed-docs] “When you make changes to a Symbol, those changes appear everywhere you use it in your designs.”
   • [claimed-docs] “Symbols can speed up your workflow by giving you a way to save and reuse common elements across your designs.”
   • [community] “Sketch is great as a design system manager. It's not all that great as a sketching tool (in spite of the name) and is a terrible photo edito…”
   • [community] “It's light, fast, cheap, easy to use and has excellent support for reusable components, making working with design systems a breeze. Adobe h…”

   proof ↗⚑ flag

Handoff

1. developerDesigns map to production code — component-to-code mappings or direct design-to-code output I can actually ship

   weight 2 · round to Framer
   Framer~partialcommunity7/10

   Framer supports genuine dev-facing code integration: Code Components (custom React components with visual controls), Code Overrides (HOCs), a Server API, and native external-agent access letting Claude Code/Cursor/Codex read/write the live project and CMS with branch-based review before merge. Community evidence corroborates that Framer's React-based model lets designers hand off working code to developers. However this is not a classic 'export design to code' pipeline (no Figma-style code export or OpenAPI-documented flow was found), and independent verification is thin/dated. missing for 10: independent hands-on proof of shipping exported production code, more recent community validation of the external-agent/dev handoff workflow.

   • [claimed-docs] “Code Components Custom React Components with visual controls to render anything on your site.”
   • [claimed-docs] “Code Overrides Higher Order Components to modify Layer and Component properties.”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [claimed-docs] “Connect Claude Code, Codex, or Cursor to update a Framer project.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, and decide when to merge …”
   • [community] “Since Framer is code based, you can do some things that are pretty amazing for a designer... Framer X is based on React, so designers can th…”

   proof ↗⚑ flag

   Sketch~partialclaimed5/10

   Sketch's AI page claims direct frame-to-code generation with '1:1 visual fidelity' and MCP-based AI client integration, plus browser-based developer handoff for inspecting specs, which together approximate design-to-code output; however this is entirely first-party marketing with no independent hands-on verification of code quality, supported frameworks, or shippability. missing for 10: independent/community confirmation that generated code is production-ready, details on output format/frameworks, and evidence of real-world dev usage of this AI code feature.

   • [claimed-docs] “Take any frame in Sketch and build it out in code with 1:1 visual fidelity.”
   • [claimed-docs] “Point your AI client at a repo to recreate text styles, colors, UI components, and even entire pages inside Sketch.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch.”
   • [claimed-docs] “they won’t even need the Mac app (or a Mac) as it all happens in a web browser”
   • [claimed-docs] “any developer can grab the information they need for any part of a design, at any time... they won't even need the Mac app (or a Mac) as it …”

   proof ↗⚑ flag

2. developerInspect any layer and get measurements, styles, assets, and code snippets (CSS/iOS/Android) without an editor seat

   weight 3 · round to Sketch
   Framer—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

   Sketch✓fullclaimed8/10

   Sketch's developer handoff docs explicitly confirm free developer invites to inspect designs in-browser without needing the Mac app or a Mac, and note developers can grab measurements/info for any part of a design at any time; code snippet generation (CSS/iOS/Android) is a longstanding Sketch inspector feature implied by 'grab the information they need' though not explicitly enumerated per-platform in this pack. missing for 10: explicit doc citation naming CSS/iOS/Android code snippet export, and independent hands-on confirmation of the inspect panel's snippet output.

   • [claimed-docs] “they won’t even need the Mac app (or a Mac) as it all happens in a web browser”
   • [claimed-docs] “You can invite developers to inspect designs in your Workspace for free”
   • [claimed-docs] “any developer can grab the information they need for any part of a design, at any time... they won't even need the Mac app (or a Mac) as it …”
   • [claimed-docs] “You can invite developers to inspect designs in your Workspace for free — they won't even need the Mac app (or a Mac) as it all happens in a…”
   • [claimed-docs] “any developer can grab the information they need for any part of a design, at any time”
   • [claimed-docs] “they won’t even need the Mac app (or a Mac) as it all happens in a web browser.”

   proof ↗⚑ flag

1. ai-native userDo everything through the API that I can do in the UI

   weight 2 · round to Framer
   Framer~partialprobed6/10

   Framer exposes a Server API, external-agent connections (Claude Code/Cursor/Codex) with read/write CMS access, branching for review, and a plugin system that covers many editor actions (framer-docs-5,17,18,24,34,35,36) — a real, substantive API surface for automation. But there's no evidence of full UI/API parity: no OpenAPI/spec discoverable (framer-probe-3 shows 404s across candidate spec paths), and the docs describe specific supported actions (CMS CRUD, canvas edits via agent, publishing) rather than a claim that literally everything achievable in the UI (e.g. fine-grained styling, animation editing, plugin marketplace management) is scriptable via API. Missing for 10: a documented comprehensive API/OpenAPI reference proving parity, explicit vendor claim of 'anything in UI = anything via API', and independent/hands-on verification of that parity.

   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Use the Server API to bring Framer into your product, or connect an external agent to work on a Framer project”
   • [claimed-docs] “Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items.”
   • [claimed-docs] “Framer doesn’t require a separate MCP server. If you’re searching for a Framer MCP, the native Framer External Agent connection gives Claude…”
   • [claimed-docs] “Yes. Once connected, the agent can read and write CMS collections, including creating, updating, and deleting items. You stay in control of …”
   • [claimed-docs] “Framer has branching, so every change made by an external agent automatically happens on a branch. You can undo changes, review the branch, …”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.framer.com/openapi.json, https://www.framer.com/swagger.json, https://www.framer.com/api…”

   proof ↗⚑ flag

   Sketch~partialprobed5/10

   Sketch exposes a JavaScript plugin API, a CLI (sketchtool), a URL scheme, and MCP-based AI integrations for asset creation and design manipulation, showing meaningful programmatic access beyond the UI. However, there's no evidence of a comprehensive REST/public API covering all UI functions (e.g., prototyping links, sharing/versioning, symbol libraries, workspace permissions) — openapi probes 404'd and docs focus on plugin/CLI/AI hooks rather than full UI parity. missing for 10: a documented full REST/GraphQL API covering all UI actions, evidence that non-plugin workflows (sharing, prototyping, library publishing) are API-accessible, and independent confirmation of API completeness.

   • [claimed-docs] “Write a Sketch plugin in JavaScript. Take advantage of ES6, access macOS frameworks and use the Sketch APIs without learning Objective-C or …”
   • [claimed-docs] “Try the Sketch JavaScript API now, directly within Sketch.”
   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”
   • [claimed-docs] “Integrate Sketch into your workflows with the `sketch://` URL scheme.”
   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://www.sketch.com/openapi.json, https://www.sketch.com/swagger.json, https://www.sketch.com/api…”

   proof ↗⚑ flag

2. ai-native userExport all of my data in open formats and leave

   weight 3 · round to Sketch
   Framer—none0/10

   Evidence shows Framer offers a Server API, CMS API, and CSV import for content, but nothing indicates the ability to export a full site (design, code, CMS data) into open, portable formats for migration away from Framer; only 'migrate to Framer' flows are documented, not exit/export paths.

   • [claimed-docs] “Added two linked collections: 1. Articles — 39 CSV rows imported. 2. Categories — 10 rows added and linked to Articles as multi-select tags.”
   • [claimed-docs] “use-cases.csv to create Use Case Articles and Categories collections. Populate fields and match each article to a category.”
   • [claimed-docs] “Server API Connect your product and automate publishing with the Framer API.”
   • [claimed-docs] “Connect your product and automate publishing with the Framer API.”

   proof ↗⚑ flag

   Sketch~partialcommunity4/10

   Sketch documents use the open `.sketch` format (a zipped JSON+binary archive) rather than a proprietary opaque format, which supports portability and third-party tooling/git integration per community reports. However, there's no evidence of a full account/workspace-level export tool for bulk data extraction (comments, version history, workspace settings) to enable a clean 'export everything and leave' migration. missing for 10: bulk workspace/account export feature, documented data-portability/migration guide, independent confirmation of successful full data export.

   • [claimed-docs] “Sketch stores documents in .sketch format, a zipped archive of JSON formatted data and binary data such as images.”
   • [claimed-docs] “Sketch stores documents in `.sketch` format, a zipped archive of JSON formatted data and binary data such as images.”
   • [community] “The file format is basically zipped json files, so there's a robust plugin ecosystem and you can do things like git integration pretty easil…”

   proof ↗⚑ flag

3. ai-native userRead the product's source under an open license

   weight 2 · round drawn
   Framer—none0/10

   Framer is closed-source proprietary SaaS; no evidence pack item mentions an open-license source availability, and this is a mainstream commercial product with no indication of open-sourcing its codebase.

   ⚑ flag

   Sketch—none0/10

   Sketch is a proprietary, closed-source commercial design tool; the file format spec (sketch-document) is open, but there is no evidence the application source code itself is under an open license. No evidence contradicts closed-source status.

   ⚑ flag

4. ai-native userSelf-host the core product

   weight 3 · round drawn
   Framer—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

   Sketch—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

Plugins

1. developerA documented plugin API with typings and samples lets me build and distribute my own editor extensions

   weight 2 · round to Sketch
   Framer~partialclaimed6/10

   Framer documents a Plugins API for building editor extensions (small apps that interact with the Editor/CMS, insert layers/images/code components) with a dedicated developer docs section, plus adjacent APIs like Code Components and Server API. However, the evidence pack doesn't show explicit TypeScript typings references or concrete code samples/SDK repo, and there's no distribution/marketplace mechanism described. missing for 10: explicit typings documentation, concrete code samples/starter templates, and a distribution/marketplace mechanism for sharing plugins.

   • [claimed-docs] “Plugins in Framer are small apps that can interact with the editor.”
   • [claimed-docs] “Plugins Build small apps that can interact with the Framer Editor and CMS.”
   • [claimed-docs] “Plugins in Framer are small apps that can interact with the editor. They enable you to perform various actions such as inserting images, lay…”
   • [claimed-docs] “Plugins in Framer are small apps that can interact with the editor. They enable you to perform various actions such as inserting images, lay…”
   • [claimed-docs] “Code Components Custom React Components with visual controls to render anything on your site.”
   • [claimed-docs] “Code Overrides Higher Order Components to modify Layer and Component properties.”

   proof ↗⚑ flag

   Sketch✓fullcommunity8/10

   Sketch provides a dedicated developer portal (developer.sketch.com) with a documented JavaScript plugin API, in-browser API tester, official vanilla JS API stable across releases, plugin manager (skpm) for publishing/distributing plugins, and manual install support, plus CLI/URL-scheme integration options. Missing for 10: independent hands-on developer reviews confirming ease of building complex plugins and richer typings/type-definition documentation beyond API reference mentions.

   • [claimed-docs] “Write a Sketch plugin in JavaScript. Take advantage of ES6, access macOS frameworks and use the Sketch APIs without learning Objective-C or …”
   • [claimed-docs] “Try the Sketch JavaScript API now, directly within Sketch.”
   • [claimed-docs] “Use the Sketch Plugin Manager `skpm` to publish a new plugin or an update of an existing one.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [claimed-docs] “you can install plugins by simply moving them into the Plugins folder yourself.”
   • [community] “The file format is basically zipped json files, so there's a robust plugin ecosystem and you can do things like git integration pretty easil…”
   • [community] “Sketch has been widely adopted for UI design. It's simpler than Adobe Illustrator, lighter in resources and subjectively feels far less clun…”

   proof ↗⚑ flag

2. designerExtend the editor from a large plugin community or marketplace — icons, data fillers, accessibility checkers, and more

   weight 2 · round to Sketch
   Framer~partialclaimed5/10

   Framer's docs confirm a plugin system where developers build 'small apps' that can interact with the editor and CMS (inserting images, layers, code components, modifying canvas), which is the technical foundation for a plugin ecosystem. However, the evidence never mentions a marketplace, community size, or specific plugin categories like icon packs, data fillers, or accessibility checkers as the story asks. Missing for 10: evidence of an actual marketplace/directory, plugin discovery UX, and named example plugins matching icons/data/accessibility use cases.

   • [claimed-docs] “Plugins in Framer are small apps that can interact with the editor.”
   • [claimed-docs] “Plugins Build small apps that can interact with the Framer Editor and CMS.”
   • [claimed-docs] “Plugins in Framer are small apps that can interact with the editor. They enable you to perform various actions such as inserting images, lay…”
   • [claimed-docs] “Plugins in Framer are small apps that can interact with the editor. They enable you to perform various actions such as inserting images, lay…”

   proof ↗⚑ flag

   Sketch~partialcommunity7/10

   Sketch provides a documented plugin API (JavaScript, ES6, macOS framework access), a CLI/manager (skpm) for publishing and installing plugins, and manual plugin installation, and community sources independently confirm Sketch has 'a massive collection of plugins that forms a large part of its appeal' and a 'robust plugin ecosystem.' However, no evidence names a curated marketplace/directory or specific plugin categories like icon libraries, data fillers, or accessibility checkers. Missing for 10: an official plugin marketplace/directory listing, explicit examples of icon/data-filler/accessibility-checker plugins, and any count or curation of the ecosystem.

   • [claimed-docs] “Write a Sketch plugin in JavaScript. Take advantage of ES6, access macOS frameworks and use the Sketch APIs without learning Objective-C or …”
   • [claimed-docs] “Use the Sketch Plugin Manager `skpm` to publish a new plugin or an update of an existing one.”
   • [claimed-docs] “you can install plugins by simply moving them into the Plugins folder yourself.”
   • [claimed-docs] “An official vanilla JavaScript API, fully supported between releases.”
   • [community] “The file format is basically zipped json files, so there's a robust plugin ecosystem and you can do things like git integration pretty easil…”
   • [community] “Sketch has been widely adopted for UI design. It's simpler than Adobe Illustrator, lighter in resources and subjectively feels far less clun…”

   proof ↗⚑ flag

Pricing

1. product managerPublished plan pricing makes seat types, tiers, and AI-feature gating clear before procurement

   weight 1 · round to Sketch
   Framer—none0/10

   The evidence pack contains no pricing page, plan tiers, seat types, or AI-feature gating details — only one mention that a static-file hosting feature is limited to Pro/Enterprise plans, which does not constitute published plan/seat/AI-gating clarity. No pricing comparison table or plan documentation is present anywhere in the pack.

   ⚑ flag

   Sketch~partialclaimed5/10

   The pricing page evidence lists enterprise-tier features like SSO, SCIM provisioning, permissions groups, and project archiving, implying a tiered structure, and sketch-docs-18 explicitly states AI/MCP features carry 'no tokens to buy, no extra tiers, no add-ons,' which speaks directly to AI-feature gating. However, there is no evidence of explicit seat-type definitions, plan names, or per-seat pricing tables that a PM could use to compare procurement options. Missing for 10: explicit seat-type/tier naming and pricing amounts, a full feature-by-tier comparison matrix, and independent confirmation that AI gating claims match published plans.

   • [claimed-docs] “Connect any MCP-compatible AI client you like to Sketch. There are no tokens to buy, no extra tiers, and no add-ons.”
   • [claimed-docs] “SCIM provisioning”
   • [claimed-docs] “Permissions groups”
   • [claimed-docs] “Single Sign-On (SSO) * Project archiving * Permissions groups”
   • [claimed-docs] “Learn how Single Sign-On (SSO) and SCIM provisioning work in Sketch, and find step-by-step guides to get set up.”

   proof ↗⚑ flag

2. designerThe free tier supports real design work — usable editor and collaboration without a paid seat

   weight 2 · round drawn
   Framer—none0/10

   The evidence pack contains no mention of Framer's free tier, its editor limitations, or free-tier collaboration features; all docs focus on AI agents, CMS, publishing, and plugins. No pricing or free-plan details are cited anywhere in the pack, so there is no basis to confirm the free tier supports real design work or collaboration.

   ⚑ flag

   Sketch—none0/10

   The evidence pack only shows free access for developers to inspect/hand-off designs in a browser (no Mac needed) and a legacy 'no forced upgrade' licensing note, but there is no evidence of a free tier that lets a designer actually create/edit designs or collaborate in real time without a paid seat — Sketch is fundamentally a purchased desktop app for designers. Community notes even highlight it lacked real-time collaboration for years and requires a Mac/license to design at all.

   • [claimed-docs] “You can invite developers to inspect designs in your Workspace for free”
   • [claimed-docs] “You can invite developers to inspect designs in your Workspace for free — they won't even need the Mac app (or a Mac) as it all happens in a…”
   • [claimed-docs] “they won’t even need the Mac app (or a Mac) as it all happens in a web browser.”
   • [community] “Newer versions of Sketch open old files just fine, and licenses are valid for any version. Sketch licenses do not 'expire', you simply can't…”
   • [community] “You don't have to pay to keep using Sketch, as long as you're ok with not upgrading. Unlike Adobe creative cloud that stops functioning the …”
   • [community] “Sketch is great software, well built, but the fact that in late 2020 they still have to write 'And soon, the ability to collaborate in real-…”

   proof ↗⚑ flag

1. ai-native userChoose where my data is stored (region/residency)

   weight 2 · round drawn
   Framer—none0/10

   No evidence in the pack addresses data residency, region selection, or data storage location controls for Framer; all evidence concerns AI agents, CMS, plugins, and publishing features unrelated to data residency.

   ⚑ flag

   Sketch—none0/10

   No evidence in the pack mentions data residency, region selection, or storage location options for Sketch's Workspace/cloud data; only SSO/SCIM/permissions security features are documented, none addressing data residency choice.

   ⚑ flag

2. ai-native userPrevent my data from being used to train AI models

   weight 3 · round drawn
   Framer—none0/10

   No evidence in the pack addresses data usage for AI training, opt-out controls, or any privacy policy statement regarding training data; all evidence concerns AI agent features, plugins, and site publishing.

   ⚑ flag

   Sketch—none0/10

   No evidence in the pack addresses AI training data opt-out, privacy policy on model training, or any data-usage controls specific to AI training; Sketch's AI features (e.g., MCP integration, image-to-palette) are documented but nothing states whether user design data is used to train models or how to opt out.

   ⚑ flag

3. ai-native userControl data retention and deletion

   weight 2 · round drawn
   Framer—none0/10

   The evidence pack covers Framer's AI agent, plugins, API, CMS, and version control/branching features, but contains no mention of data retention policies, data deletion controls, or privacy/compliance settings for AI-native workflows. Missing for 10: any documentation on data retention periods, user-initiated data deletion, or privacy controls specific to AI agent interactions.

   ⚑ flag

   Sketch—none0/10

   No evidence in the pack addresses data retention policies, deletion controls, or data export/erasure mechanisms for AI-native users; only unrelated SSO/SCIM/permissions and AI-feature docs are present.

   ⚑ flag

4. ai-native userOpt out of telemetry and usage tracking

   weight 2 · round drawn
   Framer—none0/10

   No evidence in the pack addresses telemetry, usage tracking, or an opt-out mechanism for Framer's AI features or platform; all documentation covers unrelated capabilities (agents, CMS, plugins, publishing).

   ⚑ flag

   Sketch—none0/10

   No evidence in the pack addresses telemetry, analytics, or usage-tracking opt-out settings for Sketch; the docs and community items cover features, pricing, and performance, not privacy/telemetry controls. Missing for 10: any mention of telemetry collection, opt-out settings, privacy policy details on usage data, or admin controls to disable analytics.

   ⚑ flag

1. designerPrototypes can carry advanced behavior — animation between states, conditions or variables, scroll effects — to feel like the real product

   weight 1 · round to Sketch
   Framer!disputedcontradicted6/10

   Framer's docs and community evidence confirm rich prototyping behavior — animated menus, carousels, code overrides, data-driven components (docs-45, docs-32, comm-4, comm-5) — but hands-on reports document concrete failures that undercut the 'feels like the real product' promise: animations distort during size/geometry changes, performance is inconsistent (60fps then chugging) on the same device, and users found Framer 'falls through' vs alternatives like react-spring when chaining multiple interactive animated nodes (comm-6, comm-7, comm-9). Missing for 10: independent verification that these animation glitches are resolved, and evidence of robust variable/conditional-logic support beyond CMS data binding.

   • [claimed-docs] “Sticky navbars, animated menus, interactive carousels, just describe what you want. The agent handles the wiring so you can focus on the des…”
   • [claimed-docs] “Code Overrides Higher Order Components to modify Layer and Component properties.”
   • [claimed-docs] “Fetch Building API endpoints that designers can use on their Framer sites without code.”
   • [community] “Since Framer is code based, you can do some things that are pretty amazing for a designer... Framer X is based on React, so designers can th…”
   • [community] “As a ux designer... this is quite an interesting tool if you design software. It gives a nice blend between design and code. Your prototypes…”
   • [community] “I use Framer on some projects and I love its declarative approach to creating nice animations. But one thing that has been confusing me is i…”
   • [community] “The rounded corners and the border thickness are getting distorted when animating size changes... Basically any effect that doesn't commute …”
   • [community] “I feel react-spring is better suited for the job... tried them both with r3f, i think it's just about API design and framer starts to fall t…”

   proof ↗⚑ flag

   Sketch~partialclaimed4/10

   Sketch's docs show interactive prototyping via hotspots/links, overlays, hover/press/toggle triggers, and scrolling areas, which cover some 'feel like the real product' interactions, but there is no evidence of animated transitions between states, conditional logic, or variables — capabilities competitors highlight for advanced prototyping. Missing for 10: animation/transition effects between artboards, conditional logic or variable-driven states, and any documented scroll-triggered animation beyond basic scrolling areas.

   • [claimed-docs] “Create links or Hotspots to navigate between different frames or trigger Overlays.”
   • [claimed-docs] “With features like overlays, hover, press and toggle triggers, and scrolling areas, you can accurately show how your designs would work in t…”
   • [claimed-docs] “With features like overlays, hover, press and toggle triggers, and scrolling areas, you can accurately show how your designs would work”
   • [claimed-docs] “With features like overlays, hover, press and toggle triggers, and scrolling areas, you can accurately show how your designs would work in t…”
   • [claimed-docs] “Show Overlays (e.g. menus, dropdowns or dialogs) that appear when you click, press or hover over another layer.”

   proof ↗⚑ flag

2. designerWire screens into clickable prototypes with triggers, transitions, and overlays — no code required

   weight 2 · round to Sketch
   Framer~partialcommunity4/10

   Evidence shows Framer supports animations/transitions (community: 'great for more advanced prototyping needs w/ animations/transitions', 'declarative approach to creating nice animations') and AI-driven interactive elements like 'sticky navbars, animated menus, interactive carousels' handled via prompts, but there is no explicit documentation of screen-to-screen click triggers or overlay linking (the core Figma-style prototyping flow), and community reports flag animation performance/rendering glitches that undercut confidence. missing for 10: explicit docs on multi-screen click triggers, overlay/modal linking, and independent confirmation of a smooth no-code prototyping flow.

   • [community] “Framer is great for more advanced prototyping needs w/ animations/transitions/etc, but Adobe XD is pretty good and fast for many needs, and …”
   • [community] “I use Framer on some projects and I love its declarative approach to creating nice animations. But one thing that has been confusing me is i…”
   • [community] “The rounded corners and the border thickness are getting distorted when animating size changes... Basically any effect that doesn't commute …”
   • [claimed-docs] “Sticky navbars, animated menus, interactive carousels, just describe what you want. The agent handles the wiring so you can focus on the des…”

   proof ↗⚑ flag

   Sketch✓fullclaimed8/10

   Sketch's own docs describe hotspots/links between frames, overlays (menus/dropdowns/dialogs), and hover/press/toggle triggers plus scrolling areas — directly matching the story's triggers/transitions/overlays without code, and mobile preview lets designers play prototypes on a real device. Missing for 10: independent hands-on reviews specifically praising the prototyping feature (community evidence focuses on performance/native-app debates, not prototyping quality) and no detail on transition animation styles/timing options.

   • [claimed-docs] “Create links or Hotspots to navigate between different frames or trigger Overlays.”
   • [claimed-docs] “With features like overlays, hover, press and toggle triggers, and scrolling areas, you can accurately show how your designs would work in t…”
   • [claimed-docs] “With features like overlays, hover, press and toggle triggers, and scrolling areas, you can accurately show how your designs would work”
   • [claimed-docs] “With features like overlays, hover, press and toggle triggers, and scrolling areas, you can accurately show how your designs would work in t…”
   • [claimed-docs] “Show Overlays (e.g. menus, dropdowns or dialogs) that appear when you click, press or hover over another layer.”
   • [claimed-docs] “View your documents on the go, play prototypes on a real device, and mirror designs as you edit them on your Mac.”

   proof ↗⚑ flag

Canvas

1. designerFrames can behave responsively — auto layout, constraints, and grids resize designs the way real UIs flex

   weight 2 · round to Framer
   Framer~partialclaimed6/10

   Framer's docs describe breakpoint-based responsive adaptation ('adapts your layouts across breakpoints, adjusting stacks, type sizes, spacing, and image crops') and general layout refinement (breakpoints, spacing) via its AI agent and canvas, which implies auto-layout/stack-style responsive behavior. However, there's no explicit documentation of manual constraints or grid systems, and no independent/hands-on corroboration of responsive resizing quality. missing for 10: explicit docs on constraints/grid systems, independent verification of resize fidelity across breakpoints, non-AI-agent manual workflow evidence.

   • [claimed-docs] “Ask the agent to “Make this responsive for tablet and phone.” It adapts your layouts across breakpoints, adjusting stacks, type sizes, spaci…”
   • [claimed-docs] “Start from a prompt and use the Framer canvas to refine layouts, typography, components, breakpoints, spacing, color, and effects. AI helps …”

   proof ↗⚑ flag

   Sketch—none0/10

   The evidence pack contains no mention of auto layout, resizing constraints, or layout grids for Sketch frames — it covers prototyping, symbols, plugins, AI features, and collaboration but nothing about responsive frame behavior. Missing for 10: any docs or claims about auto layout, constraint-based resizing, or grid systems that make frames responsive.

   ⚑ flag

2. designerI get professional vector tools — pen and path editing, boolean operations, precise alignment — for real interface design work

   weight 3 · round to Sketch
   Framer—none0/10

   The evidence pack is entirely about Framer's AI agents, CMS, publishing, plugins, and developer API — nothing addresses classic vector tools like pen/path editing, boolean operations, or precise alignment guides. No documentation or community evidence confirms these professional vector-editing capabilities.

   ⚑ flag

   Sketch~partialcommunity6/10

   Docs confirm non-destructive boolean operations (Union/Subtract/Intersect/Difference) and shape styling (fills, borders, shadows), core vector-editing capabilities for interface design. However, the evidence pack contains no explicit mention of pen/path tools or precise alignment/grid features, and one community comment notes Sketch's vector-editing is 'less buggy' in comparison but still implies some bugginess relative to Figma. Missing for 10: explicit documentation of pen tool/path editing, alignment/distribution tools, and independent hands-on corroboration of vector precision.

   • [claimed-docs] “Create a new shape from two or more shapes with non-destructive boolean operations — Union, Subtract, Intersect and Difference.”
   • [claimed-docs] “Apply a fill, border, drop shadow, inner shadow or bitmap layer to a shape.”
   • [community] “Figma is rather performant (to the point where it outruns Sketch in typical day-to-day design tasks), and their vector-editing is less buggy…”

   proof ↗⚑ flag

Files

1. designerExport production assets — PNG, SVG, PDF, and more — with per-layer presets, scales, and slices

   weight 2 · round to Sketch
   Framer—none0/10

   Framer is a website-builder/publishing tool, not a design/prototyping tool with layer-based asset export; the evidence pack contains no mention of PNG/SVG/PDF export, per-layer presets, scales, or slices — these are Figma/Sketch-style export features absent from Framer's docs.

   ⚑ flag

   Sketch~partialclaimed3/10

   The pack only hints at asset export via the `sketchtool` CLI ("Create assets with a single command") and never documents the actual Export panel, per-layer presets, scale factors, or slice tooling explicitly. Missing for 10: documentation of PNG/SVG/PDF format options, per-layer export presets, scale multipliers, and slice creation UI/behavior.

   • [claimed-docs] “Create assets with a single command with the Sketch command-line interface `sketchtool`.”
   • [claimed-docs] “Integrate Sketch into your workflows and server applications, on any platform.”

   proof ↗⚑ flag

2. developerThe file format is open or documented, so my designs stay portable and toolable outside the vendor's editor

   weight 1 · round to Sketch
   Framer—none0/10

   The evidence pack covers Framer's APIs, plugins, CMS, and AI agent integrations, but contains no mention of an open or documented native file format that would let designs be opened, edited, or version-controlled by external tools outside Framer's own editor/publishing pipeline. Version history, branching, and staging are all Framer-hosted features, not a portable file format.

   ⚑ flag

   Sketch~partialcommunity6/10

   Sketch documents its .sketch file format as a zipped archive of JSON and binary data (sketch-docs-23/47), and community evidence confirms this enables a plugin ecosystem and git-style tooling outside the vendor app (sketch-comm-14). However, this is not a formally published open spec/standard with versioned schema docs, and there's no independent tooling ecosystem comparable to open formats like SVG; the developer-handoff features are more about inspection than format portability. Missing for 10: a formal published format specification/schema documentation, third-party tools that read/write .sketch independent of Sketch's own SDK, and any standards-body or open-format commitment.

   • [claimed-docs] “Sketch stores documents in .sketch format, a zipped archive of JSON formatted data and binary data such as images.”
   • [claimed-docs] “Sketch stores documents in `.sketch` format, a zipped archive of JSON formatted data and binary data such as images.”
   • [community] “The file format is basically zipped json files, so there's a robust plugin ecosystem and you can do things like git integration pretty easil…”

   proof ↗⚑ flag

3. developerSelf-host the full design platform on my own infrastructure under an open-source license

   weight 2 · round drawn
   Framer—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

   Sketch—none0/10

   Sketch is a proprietary Mac-based SaaS/desktop product with no evidence of an open-source license or self-hostable infrastructure; community discussion even highlights Sketch's refusal to leave macOS or move to browser/cloud-native architecture, the opposite of self-hosting. No docs, licensing pages, or repos indicate an open-source core platform available for self-hosting.

   • [community] “It's not just their commitment to Desktop, but their commitment specifically to not move off MacOS where they have stated that they 'will no…”
   • [community] “A browser-based version of Sketch would be huge - this is Figma's killer USP at the moment.”
   • [claimed-docs] “Sketch stores documents in .sketch format, a zipped archive of JSON formatted data and binary data such as images.”

   proof ↗⚑ flag

4. designerFull version history lets me name milestones, compare, and restore any earlier state of a file

   weight 2 · round drawn
   Framer~partialclaimed5/10

   Framer's docs confirm automatic version history with one-click restore of any prior version (framer-docs-12, framer-docs-22) and branching to isolate changes (framer-docs-11, framer-docs-47), but there is no evidence of naming/labeling milestones or a side-by-side compare view between versions. Missing for 10: named milestone/checkpoint labeling, version comparison/diff view, independent hands-on confirmation of restore workflow.

   • [claimed-docs] “Version history and rollback. Every site change is saved automatically, so you can restore any version in seconds.”
   • [claimed-docs] “Every site change is saved automatically, so you can restore any version in seconds.”
   • [claimed-docs] “Branching. Explore ideas in a branch without touching the main project. Merge when it’s ready to ship.”
   • [claimed-docs] “Branches. Work on a copy of your project and keep the original clean.”

   proof ↗⚑ flag

   Sketch~partialclaimed5/10

   Sketch's docs describe 'starred versions' that let designers mark important updates in a document's history and control which versions become Library updates or are visible to Viewers/Guests, which covers naming/milestone-marking, but the evidence never describes a compare/diff view between versions or an explicit 'restore to any earlier state' UI. Missing for 10: documented version comparison/diff tooling, explicit restore-to-previous-version workflow, and any independent/hands-on confirmation of these version-history mechanics.

   • [claimed-docs] “starring a version will publish any changes as a Library update. Any unstarred versions after that won’t become Library updates.”
   • [claimed-docs] “When you star a version in a Library document, anyone using that Library will receive a Library update.”
   • [claimed-docs] “With starred versions, you can indicate important updates in a document's history, set which versions Viewers and Guests can see”
   • [claimed-docs] “With starred versions, you can indicate important updates in a document’s history, set which versions Viewers and Guests can see, and contro…”

   proof ↗⚑ flag