PayPal vs Autumn

Agent access

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

   weight 2 · round to PayPal
   PayPal✓fullprobed9/10

   Direct probe confirms a live llms.txt at developer.paypal.com/llms.txt (HTTP 200) plus .md-suffixed docs pages (e.g., api/rest.md) that agents can fetch directly, exactly matching the agent-oriented docs pattern. This is corroborated by extensive first-party agent tooling (MCP server, agent toolkit) showing PayPal actively supports agent-native consumption of its docs/APIs. Missing for 10: no independent/community verification of an agent actually using llms.txt in practice.

   • [probe] “PROBE llms.txt: HTTP 200 at https://developer.paypal.com/llms.txt # PayPal Developer Documentation > PayPal's developer documentation for i…”
   • [probe] “PROBE docs-md: HTTP 200 at https://developer.paypal.com/api/rest.md # Get started with PayPal REST APIs (/api/rest) Get started with PayPal…”
   • [claimed-docs] “PayPal built an MCP server that lets merchants use natural language with their favorite MCP clients.”

   proof ↗⚑ flag

   Autumn✓fullprobed8/10

   A probe confirms llms.txt is live at docs.useautumn.com/llms.txt returning HTTP 200 with structured doc links, directly satisfying the ability to point an agent at agent-oriented docs; this is reinforced by an official MCP server for AI assistants. Missing for 10: no independent/community confirmation of an agent successfully consuming llms.txt in practice, and no openapi.json fallback.

   • [probe] “PROBE llms.txt: HTTP 200 at https://docs.useautumn.com/llms.txt # Autumn - [Welcome to Autumn](https://docs.useautumn.com/welcome.md): Drop…”
   • [claimed-docs] “Autumn MCP connects AI assistants to Autumn's billing, customer, plan, balance, and log tools.”
   • [probe] “official MCP server documented at https://docs.useautumn.com/documentation/mcp”

   proof ↗⚑ flag

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

   weight 2 · round to PayPal
   PayPal✓fullclaimed7/10

   PayPal's core developer experience is API-first: OAuth client-credentials tokens authenticate calls without a UI, REST APIs cover orders/payments/subscriptions/invoicing/payouts, webhooks provide event-driven automation, and a sandbox environment exists for automated testing — all of which support headless/CI usage. missing for 10: explicit CI/CD pipeline examples (e.g., GitHub Actions integration), and independent hands-on confirmation of running PayPal API calls in a CI environment.

   • [claimed-docs] “Exchange your client ID and client secret for an access token. The access token authenticates your app when calling PayPal REST APIs.”
   • [claimed-docs] “Exchange your client ID and client secret for an access token.”
   • [claimed-docs] “Explore our REST API descriptions, generate code for your API clients, and import OpenAPI documents into compatible third-party tools.”
   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [claimed-docs] “The initial setup for receiving webhooks is to subscribe a listening URL for a specific app”
   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”
   • [github] “The PayPal Agent Toolkit enables popular agent frameworks... to integrate with PayPal APIs through function calling.”

   proof ↗⚑ flag

   Autumn~partialprobed5/10

   Autumn ships an `atmn` CLI that syncs pricing configs (`autumn.config.ts`) with a single command and supports non-interactive actions like `preview` and `pull`, which are scriptable and could run in CI, plus SDK calls (`check`, `track`) that are inherently headless. However, there's no explicit documentation or example of running Autumn in a CI/CD pipeline, no mention of non-interactive auth/token flags for automated environments, and no community confirmation of headless CI usage. Missing for 10: explicit CI/CD pipeline examples, non-interactive authentication guidance, and independent confirmation of automated/headless runs.

   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “You can preview your plans without pushing anything: bunx atmn preview”
   • [claimed-docs] “If you already have products created in the dashboard, run `atmn pull` to pull them into your local config.”
   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [probe] “official CLI documented at https://docs.useautumn.com/cli/getting-started”

   proof ↗⚑ flag

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

   weight 3 · round to PayPal
   PayPal✓fullprobed9/10

   PayPal ships a well-documented official MCP server in both local (npx @paypal/mcp) and remote hosted forms (mcp.sandbox.paypal.com / mcp.paypal.com) with SSE and Streamable HTTP transports, OAuth-based auth, and a broad tool inventory (orders, invoices, subscriptions, disputes, catalog, insights) usable from any MCP client; docs also confirm support for Anthropic/OpenAI LLMs. Missing for 10: independent/community hands-on validation of the MCP server itself (only vendor docs cited).

   • [claimed-docs] “PayPal built an MCP server that lets merchants use natural language with their favorite MCP clients.”
   • [claimed-docs] “PayPal ships BOTH deployment modes of its MCP server: local via `npx -y @paypal/mcp --tools=all` (PAYPAL_ACCESS_TOKEN + PAYPAL_ENVIRONMENT=S…”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”
   • [claimed-docs] “"PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Conte…”
   • [claimed-docs] “"PayPal recently launched a Model Context Protocol (MCP) server that customers can use to access the power of PayPal using natural language …”
   • [claimed-docs] “PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Contex…”
   • [probe] “official MCP server documented at https://developer.paypal.com/ai-tools/mcp-server”

   proof ↗⚑ flag

   Autumn✓fullprobed8/10

   Autumn is a billing/entitlements SaaS platform (not itself an agent), and it documents an official MCP server that connects AI assistants to its billing, customer, plan, balance, and log tools, confirmed by both docs and probe. Missing for 10: independent hands-on confirmation of the MCP server working and more detail on its tool coverage/setup.

   • [claimed-docs] “Autumn MCP connects AI assistants to Autumn's billing, customer, plan, balance, and log tools.”
   • [probe] “official MCP server documented at https://docs.useautumn.com/documentation/mcp”

   proof ↗⚑ flag

4. ai-native userUse an official CLI

   weight 2 · round to Autumn
   PayPal—none0/10

   Evidence covers PayPal's REST APIs, SDKs, MCP server, and agent toolkit, but no official CLI tool is mentioned anywhere in the docs or GitHub references. Missing for 10: any documented PayPal CLI, command-line tool, or terminal-based interface for developers/agents.

   ⚑ flag

   Autumn✓fullprobed8/10

   Autumn ships an official `atmn` CLI documented in first-party docs, supporting config-as-code (autumn.config.ts), pushing/syncing plans, previewing changes (bunx atmn preview), and pulling existing dashboard products (atmn pull) — clearly enabling AI-native, code-first workflows. Missing for 10: independent/hands-on community verification of the CLI itself (community evidence covers billing/security topics, not CLI usage) and no evidence of scripting/automation-specific features beyond basic commands.

   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “You can preview your plans without pushing anything: bunx atmn preview”
   • [claimed-docs] “If you already have products created in the dashboard, run `atmn pull` to pull them into your local config.”
   • [probe] “official CLI documented at https://docs.useautumn.com/cli/getting-started”

   proof ↗⚑ flag

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

   weight 3 · round to PayPal
   PayPal✓fullclaimed9/10

   PayPal has extensive, well-documented public REST APIs (OAuth token exchange, checkout, subscriptions, invoicing, payouts, disputes, webhooks) plus an OpenAPI spec/codegen tooling and a sandbox environment, all directly consumable by an AI-native user or agent. It also ships an official MCP server (local and remote), an agent toolkit for LangChain/CrewAI/OpenAI Agents SDK/Bedrock, and documented agentic commerce protocols (ACP/UCP), showing the API is explicitly built for agent-driven use. Missing for 10: independent/third-party developer corroboration of API reliability or friction beyond PayPal's own docs.

   • [claimed-docs] “Exchange your client ID and client secret for an access token. The access token authenticates your app when calling PayPal REST APIs.”
   • [claimed-docs] “Explore our REST API descriptions, generate code for your API clients, and import OpenAPI documents into compatible third-party tools.”
   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”
   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [github] “The PayPal Agent Toolkit enables popular agent frameworks... to integrate with PayPal APIs through function calling.”
   • [claimed-docs] “PayPal ships BOTH deployment modes of its MCP server: local via `npx -y @paypal/mcp --tools=all` (PAYPAL_ACCESS_TOKEN + PAYPAL_ENVIRONMENT=S…”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”
   • [claimed-docs] “"PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Conte…”
   • [claimed-docs] “"Agent Ready helps Braintree merchants accept payments from AI shopping assistants across major platforms, including ChatGPT, Google AI Mode…”

   proof ↗⚑ flag

   Autumn✓fullprobed8/10

   Autumn ships a documented public API with clear reference pages (check, track, attach) and code examples, webhooks, and an API-reference section, letting an AI-native user drive billing/entitlements programmatically. Missing for 10: a discoverable machine-readable OpenAPI/swagger spec (probe found only 404s at expected paths), and independent hands-on confirmation of API robustness beyond docs.

   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [claimed-docs] “You can also [check and reserve] balance in a single, atomic API call for concurrent events.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [claimed-docs] “Fired whenever a customer's plans change — new subscriptions, upgrades, downgrades, etc.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.useautumn.com/openapi.json, https://docs.useautumn.com/swagger.json, https://docs.useau…”

   proof ↗⚑ flag

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

   weight 2 · round drawn
   PayPal—none0/10

   Evidence shows only standard OAuth client-credentials token exchange (client ID/secret → access token) and general agent-toolkit/MCP integration, but nothing documenting scoped, restricted, or least-privilege credential issuance specifically for agent use (e.g., limited-permission API keys, granular scopes, or per-agent restricted tokens).

   • [claimed-docs] “Exchange your client ID and client secret for an access token. The access token authenticates your app when calling PayPal REST APIs.”
   • [claimed-docs] “Exchange your client ID and client secret for an access token.”
   • [claimed-docs] “PayPal ships BOTH deployment modes of its MCP server: local via `npx -y @paypal/mcp --tools=all` (PAYPAL_ACCESS_TOKEN + PAYPAL_ENVIRONMENT=S…”
   • [claimed-docs] “"PayPal recently launched a Model Context Protocol (MCP) server that customers can use to access the power of PayPal using natural language …”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn's docs describe customerId-based check/track/attach calls and an MCP server for billing tools, but there is no evidence of an API-key or credential system that lets an AI-native user mint scoped, least-privilege credentials for an agent. Community reports even describe unauthenticated server actions and customer_id spoofing (IDOR), the opposite of least-privilege access control.

   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [claimed-docs] “Autumn MCP connects AI assistants to Autumn's billing, customer, plan, balance, and log tools.”
   • [community] “Commenters criticized Autumn's 'backendless billing' approach as reinventing JWTs poorly: "Your encrypted customer ID has the exact same sem…”
   • [community] “Security concern raised: Autumn's server actions were public, unauthenticated routes, meaning "a simple curl request with a different custom…”

   proof ↗⚑ flag

7. ai-native userBuild against official SDKs

   weight 2 · round to PayPal
   PayPal✓fullclaimed8/10

   PayPal ships official REST API SDKs, OpenAPI specs with codegen, and a dedicated open-source Agent Toolkit (GitHub) supporting LangChain, OpenAI Agents SDK, CrewAI, Vercel AI SDK, Bedrock, and MCP for building AI-native payment workflows, backed by sandbox testing and detailed tool docs. Missing for 10: independent/hands-on developer corroboration of SDK quality beyond first-party docs and GitHub README.

   • [github] “The PayPal Agent Toolkit enables popular agent frameworks... to integrate with PayPal APIs through function calling.”
   • [github] “The PayPal Agent Toolkit enables popular agent frameworks including OpenAI's Agent SDK, LangChain, Vercel's AI SDK, and Model Context Protoc…”
   • [claimed-docs] “PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Contex…”
   • [claimed-docs] “"PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Conte…”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”
   • [claimed-docs] “Explore our REST API descriptions, generate code for your API clients, and import OpenAPI documents into compatible third-party tools.”
   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”
   • [claimed-docs] “npm run codegen-java -- openapi/checkout_orders_v2.json --artifact-id orders”

   proof ↗⚑ flag

   Autumn✓fullprobed7/10

   Docs consistently show official SDK code (autumn.check, autumn.track, autumn.attach) across many endpoints, plus a companion CLI (atmn) for config sync, indicating a first-party SDK for building AI-native billing integrations. Missing for 10: explicit multi-language SDK coverage/package registry links and independent developer corroboration of SDK usage beyond docs.

   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [probe] “official CLI documented at https://docs.useautumn.com/cli/getting-started”

   proof ↗⚑ flag

8. ai-native userSubscribe to events via webhooks

   weight 2 · round to PayPal
   PayPal✓fullclaimed8/10

   PayPal's REST API documents webhooks as HTTPS posts to a subscribed endpoint whenever a specified event occurs, with a clear subscription setup flow (paypal-docs-9, paypal-docs-28). This is a standard developer-facing capability well documented for programmatic/agentic integration. Missing for 10: no independent/hands-on corroboration of webhook reliability or delivery guarantees, and no explicit mention of AI-agent-specific webhook consumption patterns.

   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [claimed-docs] “The initial setup for receiving webhooks is to subscribe a listening URL for a specific app”

   proof ↗⚑ flag

   Autumn~partialclaimed6/10

   Autumn documents webhooks fired on plan/subscription changes (autumn-docs-8), giving basic event subscription capability, but evidence doesn't cover a broad event catalog (e.g. usage/track events), payload schemas, signing/verification, or subscription management API — missing for 10: full webhook event catalog, payload/signature docs, subscription management endpoints, independent hands-on confirmation.

   • [claimed-docs] “Fired whenever a customer's plans change — new subscriptions, upgrades, downgrades, etc.”

   proof ↗⚑ flag

Agentic features

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

   weight 2 · round to PayPal
   PayPal✓fullclaimed8/10

   PayPal ships an official MCP server (local and remote) plus an Agent Toolkit supporting multiple agent frameworks, letting users/agents issue natural-language commands to create orders, invoices, subscriptions, refunds, and get merchant insights, and also supports agentic commerce protocols (ACP/UCP) for conversational checkout. missing for 10: independent/hands-on user reports confirming reliability of natural-language command execution, and some commerce tools are gated behind access requests or feature flags (gift-cards-only, limited to US physical goods).

   • [claimed-docs] “PayPal built an MCP server that lets merchants use natural language with their favorite MCP clients.”
   • [github] “The PayPal Agent Toolkit enables popular agent frameworks... to integrate with PayPal APIs through function calling.”
   • [github] “The PayPal Agent Toolkit enables popular agent frameworks including OpenAI's Agent SDK, LangChain, Vercel's AI SDK, and Model Context Protoc…”
   • [claimed-docs] “PayPal ships BOTH deployment modes of its MCP server: local via `npx -y @paypal/mcp --tools=all` (PAYPAL_ACCESS_TOKEN + PAYPAL_ENVIRONMENT=S…”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”
   • [claimed-docs] “"PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Conte…”
   • [claimed-docs] “"PayPal recently launched a Model Context Protocol (MCP) server that customers can use to access the power of PayPal using natural language …”
   • [claimed-docs] “"Merchants can use PayPal's agentic commerce services to create AI-powered shopping experiences, so customers can shop using everyday langua…”
   • [claimed-docs] “Store Sync "connects your product catalog and commerce API with PayPal's agentic commerce services, enabling AI agents to discover your prod…”
   • [claimed-docs] “PayPal's remote MCP server now supports large language models (LLMs) from Anthropic and OpenAI, which brings even greater flexibility.”

   proof ↗⚑ flag

   Autumn~partialprobed5/10

   Autumn ships an official MCP server that 'connects AI assistants to Autumn's billing, customer, plan, balance, and log tools,' enabling AI-native/natural-language operation indirectly through an MCP-connected assistant rather than a built-in chat interface. There's no first-party or community evidence of actual natural-language command usage in practice, and the product is fundamentally a CLI/API/dashboard tool otherwise. Missing for 10: direct natural-language command interface (not just MCP passthrough), hands-on demonstration of NL commands working, independent user corroboration.

   • [claimed-docs] “Autumn MCP connects AI assistants to Autumn's billing, customer, plan, balance, and log tools.”
   • [probe] “official MCP server documented at https://docs.useautumn.com/documentation/mcp”

   proof ↗⚑ flag

Api quality

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

   weight 2 · round to PayPal
   PayPal~partialclaimed5/10

   PayPal's docs let you generate client code and import OpenAPI specs, and Braintree's GraphQL developer hub explicitly offers 'an in-browser API Explorer for testing queries' with runnable example integrations — concrete evidence of an interactive, runnable API reference for at least one product line. However, no evidence shows an equivalent try-it-now/runnable console for the core PayPal REST APIs (orders, checkout, subscriptions, etc.) themselves. Missing for 10: in-browser runnable examples for the main REST API reference, independent/hands-on confirmation of the interactive experience quality.

   • [claimed-docs] “Explore our REST API descriptions, generate code for your API clients, and import OpenAPI documents into compatible third-party tools.”
   • [claimed-docs] “npm run codegen-java -- openapi/checkout_orders_v2.json --artifact-id orders”
   • [claimed-docs] “Braintree (PayPal's gateway) exposes a first-class GraphQL API — "Resources and tools to integrate with Braintree's GraphQL API" — with guid…”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn's docs include static code snippets for API endpoints (e.g. check, attach) but there is no evidence of an interactive, runnable API console—probes for openapi/swagger specs all 404'd, indicating no interactive reference tooling exists.

   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.useautumn.com/openapi.json, https://docs.useautumn.com/swagger.json, https://docs.useau…”

   proof ↗⚑ flag

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

   weight 2 · round to PayPal
   PayPal✓fullclaimed9/10

   PayPal explicitly documents downloadable/importable OpenAPI specs: 'Explore our REST API descriptions, generate code for your API clients, and import OpenAPI documents into compatible third-party tools' and provides a GitHub repo (paypal-rest-api-specifications) with actual OpenAPI JSON files used for codegen. Missing for 10: no independent third-party confirmation of spec completeness/versioning outside PayPal's own repo.

   • [claimed-docs] “Explore our REST API descriptions, generate code for your API clients, and import OpenAPI documents into compatible third-party tools.”
   • [claimed-docs] “npm run codegen-java -- openapi/checkout_orders_v2.json --artifact-id orders”

   proof ↗⚑ flag

   Autumn—none0/10

   There's an api-reference section implying documented endpoints, but a direct probe for OpenAPI/Swagger specs at standard paths returned 404s with no alternate machine-readable spec found or linked anywhere in the docs.

   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.useautumn.com/openapi.json, https://docs.useautumn.com/swagger.json, https://docs.useau…”

   proof ↗⚑ flag

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

   weight 1 · round to PayPal
   PayPal✓fullclaimed8/10

   PayPal documents a dedicated sandbox as 'a self-contained, virtual testing environment that simulates the live PayPal production environment' (paypal-docs-11), and the agent toolkit explicitly instructs AI-native builders to 'Always use the sandbox environment for initial testing to avoid real transactions' (paypal-supp-toolkit-frameworks), with both local and remote MCP servers supporting PAYPAL_ENVIRONMENT=SANDBOX and a distinct sandbox endpoint (mcp.sandbox.paypal.com) separate from production (paypal-supp-remote-mcp). missing for 10: independent/hands-on confirmation that sandbox behavior fully mirrors production for agentic/AI workflows, and more detail on how sandbox data isolation is verified beyond vendor docs.

   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”
   • [claimed-docs] “"PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Conte…”
   • [claimed-docs] “PayPal ships BOTH deployment modes of its MCP server: local via `npx -y @paypal/mcp --tools=all` (PAYPAL_ACCESS_TOKEN + PAYPAL_ENVIRONMENT=S…”
   • [claimed-docs] “PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Contex…”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn is a billing/entitlements layer where sandbox/test-mode testing is a reasonable ask (similar to Stripe test mode), but the evidence pack has no mention of a dedicated sandbox/test environment separate from production—only 'atmn preview' for previewing pricing plan changes before pushing config, which is not the same as isolating test transactions/customers from production data.

   • [claimed-docs] “You can preview your plans without pushing anything: bunx atmn preview”

   proof ↗⚑ flag

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

   weight 2 · round drawn
   PayPal—none0/10

   Evidence shows PayPal APIs use version numbers (e.g., 'checkout_orders_v2.json') and OpenAPI specs, but no evidence pack item documents a formal deprecation policy, sunset timeline, or versioning support commitment that an AI-native integrator could rely on.

   • [claimed-docs] “npm run codegen-java -- openapi/checkout_orders_v2.json --artifact-id orders”

   proof ↗⚑ flag

   Autumn—none0/10

   No evidence of API versioning scheme or documented deprecation policy; changelog entries show ongoing changes but no version numbers or deprecation guarantees, and OpenAPI/spec probes 404. missing for 10: versioned API endpoints/headers, explicit deprecation policy documentation, migration guides for breaking changes.

   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.useautumn.com/openapi.json, https://docs.useautumn.com/swagger.json, https://docs.useau…”
   • [claimed-docs] “Usage alerts gain a `basis` field that defines what 100% means: the full `balance` (the default, unchanged), the plan’s `included` grants, `…”
   • [claimed-docs] “Catalog plans now accept a `processors` object so Autumn can reuse your existing Stripe catalog instead of minting new objects.”

   proof ↗⚑ flag

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

   weight 2 · round to PayPal
   PayPal~partialclaimed6/10

   PayPal's Payouts API explicitly supports true bulk operations — sending money to 'thousands of recipients at once' via email, phone, or PayPal ID in a single batch call — which is a strong general REST-API bulk capability accessible to automation. However, the AI-agent-specific surface (MCP server / Agent Toolkit tool inventory) exposes only single-item tools (create_order, get_invoice, list_transaction, etc.) with no documented batch/bulk tool for an AI agent to invoke across many items at once. Missing for 10: an AI-native (MCP/agent-toolkit) tool that explicitly performs bulk/batch actions, and any hands-on evidence of an agent successfully driving the Payouts batch API end-to-end.

   • [claimed-docs] “Send money immediately to multiple recipients with their emails, phone numbers, or PayPal IDs.”
   • [claimed-docs] “Send money to thousands of recipients at once for vendors and disbursements in various currencies.”
   • [claimed-docs] “Send money to thousands of recipients at once for vendors and disbursements in various currencies... A basic solution for sending batch paym…”
   • [claimed-docs] “A basic solution for sending batch payments using recipients' email addresses, phone numbers, or PayPal IDs.”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn's docs show only single-item operations (check, track, attach for one customer/feature at a time) and CLI commands (pull/preview/push) that sync a single config, but no batch/bulk API for operating across many customers, features, or items at once. missing for 10: bulk/batch API endpoints, documentation of batch customer or feature updates, evidence of any multi-item transaction support.

   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”

   proof ↗⚑ flag

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

   weight 3 · round drawn
   PayPal~partialclaimed4/10

   PayPal exposes webhooks that fire on transaction/dispute/subscription events (paypal-docs-9, paypal-docs-28) and has some built-in automated behaviors like automated payment recovery for failed subscriptions (paypal-docs-40) and a Disputes API for automating dispute actions (paypal-docs-26), which together give developers the raw event/trigger primitives needed to build 'if event then action' automation. However, there is no evidence of a declarative rules interface where an AI-native user defines conditions and actions without writing custom server-side handling code — webhooks require the developer to build the listener and action logic themselves. Missing for 10: a no-code/low-code rule-definition UI or API, support for arbitrary conditional logic across events, and evidence that AI agents can create/manage such rules via the MCP/agent toolkit (which only exposes discrete action tools, not rule/trigger definitions).

   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [claimed-docs] “The initial setup for receiving webhooks is to subscribe a listening URL for a specific app”
   • [claimed-docs] “Disputes API: A programmatic interface that enables you to automate dispute management actions directly from your systems.”
   • [claimed-docs] “You can customize a subscriptions integration to: Create plans that charge users a fixed amount at regular intervals or a variable amount ba…”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”

   proof ↗⚑ flag

   Autumn~partialclaimed4/10

   Autumn supports event-driven mechanisms like webhooks fired on plan changes and usage alerts with configurable thresholds (basis field), which act as simple automated triggers tied to billing state. However, there is no evidence of a general-purpose rule engine where users can define custom trigger→action logic across arbitrary events; the automation is scoped narrowly to billing/entitlement events. Missing for 10: a documented rules/automation DSL or workflow builder, support for arbitrary custom triggers/actions beyond billing events, and evidence of chaining multiple actions per rule.

   • [claimed-docs] “Fired whenever a customer's plans change — new subscriptions, upgrades, downgrades, etc.”
   • [claimed-docs] “Usage alerts gain a `basis` field that defines what 100% means: the full `balance` (the default, unchanged), the plan’s `included` grants, `…”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [claimed-docs] “You can also [check and reserve] balance in a single, atomic API call for concurrent events.”

   proof ↗⚑ flag

Ai billing ops

1. ai-native userHave an agent monitor failed payments via API and webhooks and drive retries and dunning outreach

   weight 3 · round to PayPal
   PayPal~partialclaimed5/10

   PayPal's subscriptions API documents 'Automate payment recovery for failed payments' (dunning) and webhooks can notify apps of billing failures, and the agent-toolkit/MCP tools expose subscription and invoice management functions an agent could call. However, there's no documented dedicated 'failed payment' webhook-driven retry workflow guide, no specific dunning/outreach tooling in the agent-tools inventory (only generic subscription create/update/cancel), and no evidence of an agent specifically monitoring failed payments end-to-end. missing for 10: explicit failed-payment webhook event documentation, a dedicated retry/dunning API or agent tool, and a concrete example of an agent autonomously handling failed payment recovery.

   • [claimed-docs] “You can customize a subscriptions integration to: Create plans that charge users a fixed amount at regular intervals or a variable amount ba…”
   • [claimed-docs] “Create customizable PayPal subscription plans for regular billing, offering trials, upgrades, and automated payment recovery for flexible cu…”
   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [claimed-docs] “The initial setup for receiving webhooks is to subscribe a listening URL for a specific app”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”
   • [github] “The PayPal Agent Toolkit enables popular agent frameworks... to integrate with PayPal APIs through function calling.”

   proof ↗⚑ flag

   Autumn—none0/10

   Evidence shows Autumn webhooks fire only on plan changes (new subscriptions, upgrades, downgrades) and covers usage/credit tracking, but there is no mention of failed-payment events, retry logic, or dunning outreach workflows anywhere in the docs, changelog, or community threads. This is a fair axis for a billing product (Stripe-based dunning is a common billing capability) but no evidence supports Autumn delivering it.

   • [claimed-docs] “Fired whenever a customer's plans change — new subscriptions, upgrades, downgrades, etc.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [claimed-docs] “Usage alerts gain a `basis` field that defines what 100% means: the full `balance` (the default, unchanged), the plan’s `included` grants, `…”

   proof ↗⚑ flag

2. ai-native userHave an agent provision a new product, price, and payment link via the API in a single session

   weight 3 · round to PayPal
   PayPal✓fullclaimed8/10

   PayPal's agent-toolkit/MCP tool inventory explicitly includes create_product, and subscription/plan creation tools (create plan, create_order/pay_order) plus payment-link style checkout order creation, all callable via API/function-calling in one agent session with sandbox testing supported. missing for 10: no single documented end-to-end walkthrough showing product+price+payment-link created in one session, and no independent/hands-on report confirming this exact multi-step flow works in practice.

   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”
   • [claimed-docs] “create_product: Create a new product in the PayPal catalog.”
   • [claimed-docs] “`create_product`: Create a new product in the PayPal catalog.”
   • [claimed-docs] “"PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Conte…”
   • [claimed-docs] “Create plans that charge users a fixed amount at regular intervals or a variable amount based on the number of users subscribed.”
   • [claimed-docs] “Swiftly create a Payment Link to share on your website, via email, on social media, or through text. No coding skills required.”
   • [github] “The PayPal Agent Toolkit enables popular agent frameworks including OpenAI's Agent SDK, LangChain, Vercel's AI SDK, and Model Context Protoc…”

   proof ↗⚑ flag

   Autumn~partialprobed5/10

   Autumn's API/CLI supports creating products/plans (attach, config.ts, atmn CLI) and MCP server exposes billing/plan/customer tools to AI assistants, suggesting an agent could provision products and prices programmatically, but there is no direct evidence of generating a 'payment link' via API or of a documented single-session workflow combining product+price+payment link creation. missing for 10: explicit payment-link creation endpoint/API, end-to-end example of an agent provisioning product+price+payment link in one session, independent verification of this workflow.

   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [claimed-docs] “Autumn MCP connects AI assistants to Autumn's billing, customer, plan, balance, and log tools.”
   • [probe] “official MCP server documented at https://docs.useautumn.com/documentation/mcp”
   • [probe] “official CLI documented at https://docs.useautumn.com/cli/getting-started”

   proof ↗⚑ flag

Subscriptions

1. finance-ops userAutomatically retry failed subscription payments with smart dunning and recovery emails

   weight 2 · round to PayPal
   PayPal~partialclaimed5/10

   PayPal's Subscriptions docs explicitly list 'Automate payment recovery for failed payments' as a built-in capability, directly matching the dunning/retry story, but the evidence is a single bullet point with no detail on retry schedules, dunning logic, or recovery email customization/branding. Invoicing docs also mention 'automatic payment reminders' but that's for invoices, not subscription dunning specifically. missing for 10: detailed documentation of retry cadence/logic, evidence of customizable recovery email templates, and independent/hands-on confirmation that dunning actually reduces churn.

   • [claimed-docs] “You can customize a subscriptions integration to: Create plans that charge users a fixed amount at regular intervals or a variable amount ba…”
   • [claimed-docs] “Create customizable PayPal subscription plans for regular billing, offering trials, upgrades, and automated payment recovery for flexible cu…”
   • [claimed-docs] “Merchants can share invoices via a secure URL, merchant-branded email, or scannable QR code, and configure automatic payment reminders.”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn's evidence covers metering, entitlements, plan attachment, and Stripe catalog sync, but there is no mention of dunning, retry logic for failed payments, or recovery emails anywhere in the docs, changelog, or community discussion.

   ⚑ flag

2. developerSet up recurring subscription billing with plans, trials, upgrades, and proration

   weight 3 · round to PayPal
   PayPal✓fullclaimed9/10

   PayPal's Subscriptions API docs explicitly cover creating fixed/variable billing plans, free/discounted trials, upgrade/downgrade of plans, and automated payment recovery for failed payments — covering plans, trials, upgrades and proration-like variable billing. Dashboard management and REST API integration paths are both documented. Missing for 10: no independent/hands-on third-party corroboration of proration mechanics or edge-case billing behavior.

   • [claimed-docs] “Offer your subscribers free or discounted trials... Enable subscribers to upgrade or downgrade their plans.”
   • [claimed-docs] “Create plans that charge users a fixed amount at regular intervals or a variable amount based on the number of users subscribed.”
   • [claimed-docs] “Offer your subscribers free or discounted trials.”
   • [claimed-docs] “You can customize a subscriptions integration to: Create plans that charge users a fixed amount at regular intervals or a variable amount ba…”
   • [claimed-docs] “Create customizable PayPal subscription plans for regular billing, offering trials, upgrades, and automated payment recovery for flexible cu…”
   • [claimed-docs] “Manage Subscriptions on your account dashboard... Recommended if you don't need to integrate Subscriptions into your product UI.”
   • [claimed-docs] “Manage Subscriptions on your account dashboard - Recommended if you don't need to integrate Subscriptions into your product UI.”

   proof ↗⚑ flag

   Autumn✓fullcommunity8/10

   Autumn's docs explicitly cover defining plans (autumn.config.ts + atmn CLI), attaching plans to customers with automatic handling of new subscriptions, upgrades, and downgrades, usage-based/pay-per-use pricing with proration-like billing period logic, and webhooks firing on plan changes (new subscriptions, upgrades, downgrades). Community feedback corroborates real-world use for billing/entitlements and notes Autumn's team handling of annual/monthly billing transitions and race conditions. Missing for 10: explicit first-party documentation of free trials and detailed proration mechanics, plus independent hands-on verification of trial/upgrade billing flows.

   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [claimed-docs] “Fired whenever a customer's plans change — new subscriptions, upgrades, downgrades, etc.”
   • [claimed-docs] “Pay-per-use (usage-based) pricing charges customers based on how much of a feature they actually consume, billed at the end of each billing …”
   • [claimed-docs] “Catalog plans now accept a `processors` object so Autumn can reuse your existing Stripe catalog instead of minting new objects.”
   • [community] “Autumn combines billing, metering, and entitlements into a single state, unlike Lago which only handles billing—users must sync permissions …”
   • [community] “"love this as someone who's been fixing the same billing bugs over and over and who sometimes finds stripe more complex than it should be." …”

   proof ↗⚑ flag

Card acceptance

1. founderAccept an online card payment within a day of signing up

   weight 3 · round to PayPal
   PayPal✓fullcommunity7/10

   PayPal explicitly markets no-code, fast paths to accept card payments — Payment Links ('swiftly create... no coding skills required'), PayPal Checkout buttons, and card field support — alongside sandbox testing and REST APIs for those wanting custom integration, all consistent with a founder starting to accept card payments quickly after signup. No independent hands-on report confirms the exact 'within a day' timeline, and community threads flag post-signup fund holds/verification friction that could affect real-world speed, though none concretely refute the initial-setup speed claim. Missing for 10: independent/hands-on confirmation of same-day account approval and first live transaction, and documentation of KYC/verification turnaround times.

   • [claimed-docs] “Swiftly create a Payment Link to share on your website, via email, on social media, or through text. No coding skills required.”
   • [claimed-docs] “Offer up to 400M users1 the choice to pay with PayPal, Venmo, Pay Later, debit and credit cards, and more.”
   • [claimed-docs] “Accept PayPal payments on your website. Customers can select the PayPal button on your site and pay with any funding source linked to their …”
   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”
   • [claimed-docs] “Automatically recognize guest shoppers with Fastlane by PayPal. Next time they shop, we’ll autofill their checkout credentials.”
   • [community] “After two eBay/PayPal sales cycles, PayPal emailed: 'money from payments you receive will be placed in a pending balance for up to 21 days..…”

   proof ↗⚑ flag

   Autumn~partialcommunity4/10

   Autumn is a billing/metering layer built on top of Stripe, with an `attach` API that handles new subscriptions/upgrades/downgrades and a CLI that syncs pricing plans in a single command, suggesting reasonably fast setup. However, there's no direct evidence of a turnkey checkout page or explicit same-day payment acceptance flow, and the product only works atop Stripe (comm-4), meaning a founder still needs a separate Stripe account/setup for actual card processing. Missing for 10: an end-to-end checkout UI or hosted payment page, explicit time-to-first-payment claims, and independent confirmation of same-day integration speed.

   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [community] “Concern raised that Autumn is only built on Stripe, with no support for other payment providers like Paystack, which is needed for regions l…”

   proof ↗⚑ flag

Checkout surfaces

1. developerEmbed customizable payment components directly in my own site or app instead of redirecting to a hosted page

   weight 3 · round to PayPal
   PayPal~partialclaimed6/10

   PayPal Checkout docs confirm embeddable buttons and card fields ('card field customization with your business branding') plus server-side endpoints for refunds/voids/captures, allowing developers to embed payment UI without full redirect to a hosted PayPal page. However, evidence leans heavily on button/card-field embedding rather than deep low-level UI customization (e.g., fully styled custom checkout forms), and PayPal's classic checkout flow still often redirects/pops up an approval window rather than a fully inline, brandless flow. missing for 10: detailed documentation on styling/theming APIs for embedded components, independent/hands-on confirmation that redirect can be fully eliminated, and clarity on which products (Checkout vs Braintree) offer the most customizable embedded UI.

   • [claimed-docs] “Customers can select the PayPal button on your site and pay with any funding source linked to their PayPal account.”
   • [claimed-docs] “Accept PayPal payments on your website. Customers can select the PayPal button on your site and pay with any funding source linked to their …”
   • [claimed-docs] “Expanded Checkout includes all offerings from PayPal Checkout and card field customization with your business branding.”
   • [claimed-docs] “Add server-side endpoints to handle refunds, voids, delayed captures, and specialized checkout flows.”
   • [claimed-docs] “Delay capture (/checkout/delay-capture): Charge after service delivery.”
   • [claimed-docs] “Void an authorization (/checkout/void-authorized-payment): Cancel holds before charging.”
   • [claimed-docs] “Automatically recognize guest shoppers with Fastlane by PayPal. Next time they shop, we’ll autofill their checkout credentials.”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn's evidence describes a billing/metering/entitlements API, CLI, and MCP integration built on top of Stripe, but nothing in the pack mentions embeddable payment/checkout UI components as an alternative to hosted checkout pages—attach/billing calls appear to route through Stripe's own flows.

   ⚑ flag

2. developerRedirect customers to a vendor-hosted, conversion-optimized checkout page without building my own payment UI

   weight 3 · round to PayPal
   PayPal✓fullclaimed9/10

   PayPal Checkout provides a hosted PayPal button/payment page where customers pay via their PayPal-linked funding sources, and Invoicing explicitly offers 'a secure PayPal payment page, without building their own payment flow' — both are vendor-hosted, conversion-optimized checkout surfaces requiring only redirect/embed integration plus server-side endpoints for capture/void/refund. Missing for 10: no independent conversion-rate benchmarking or hands-on developer review confirming the 'conversion-optimized' claim beyond PayPal's own marketing copy.

   • [claimed-docs] “Customers can select the PayPal button on your site and pay with any funding source linked to their PayPal account.”
   • [claimed-docs] “Accept PayPal payments on your website. Customers can select the PayPal button on your site and pay with any funding source linked to their …”
   • [claimed-docs] “Add server-side endpoints to handle refunds, voids, delayed captures, and specialized checkout flows.”
   • [claimed-docs] “PayPal Invoicing lets merchants and partners create and send itemized invoices and get paid through a secure PayPal payment page, without bu…”
   • [claimed-docs] “Automatically recognize guest shoppers with Fastlane by PayPal. Next time they shop, we’ll autofill their checkout credentials.”
   • [claimed-docs] “Delay capture (/checkout/delay-capture): Charge after service delivery.”
   • [claimed-docs] “Void an authorization (/checkout/void-authorized-payment): Cancel holds before charging.”

   proof ↗⚑ flag

   Autumn—none0/10

   The evidence pack covers Autumn's metering, entitlements, CLI, MCP, and plan-attach APIs, but nowhere describes a vendor-hosted, conversion-optimized checkout page that a developer can redirect customers to; 'attach' only documents subscription state changes, not a hosted checkout flow.

   ⚑ flag

No code selling

1. founderCreate a shareable payment link from the dashboard with no code and start selling immediately

   weight 2 · round to PayPal
   PayPal✓fullclaimed9/10

   PayPal's business site directly documents no-code Payment Links: 'Swiftly create a Payment Link to share on your website, via email, on social media, or through text. No coding skills required,' alongside broad acceptance of PayPal, Venmo, cards, and Pay Later. This directly matches the founder story of creating a shareable link with no code and selling immediately. Missing for 10: independent hands-on confirmation of the payment-link creation flow beyond vendor docs.

   • [claimed-docs] “Swiftly create a Payment Link to share on your website, via email, on social media, or through text. No coding skills required.”
   • [claimed-docs] “Offer up to 400M users1 the choice to pay with PayPal, Venmo, Pay Later, debit and credit cards, and more.”
   • [claimed-docs] “PayPal Open helps you accept payments online, in person, and on the go.”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn is a code-first billing/entitlements layer requiring SDK calls, an autumn.config.ts file, or CLI (atmn) to define and attach plans—there is no evidence of a no-code, dashboard-generated shareable payment link feature for founders to start selling immediately.

   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “If you already have products created in the dashboard, run `atmn pull` to pull them into your local config.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [claimed-docs] “Pay-per-use (usage-based) pricing charges customers based on how much of a feature they actually consume, billed at the end of each billing …”

   proof ↗⚑ flag

Payment methods

1. founderOffer digital wallets and local payment methods (Apple Pay, Google Pay, bank redirects, BNPL) from the same integration

   weight 2 · round to PayPal
   PayPal✓fullclaimed8/10

   PayPal Checkout explicitly bundles wallet funding sources (any linked PayPal funding source), Pay in 4 BNPL, and in-store tap-to-pay with Apple Pay/Google Pay/Samsung Pay via a single PayPal Open/Checkout integration, plus Expanded Checkout adds card fields and branding under the same account. This covers wallets, BNPL, and digital wallet acceptance from one integration, though bank redirect methods (e.g., iDEAL, SEPA, giropay) and full country-by-country local payment method coverage aren't explicitly enumerated in evidence. missing for 10: explicit documentation of bank-redirect/local APM coverage (iDEAL, SOFORT, giropay, etc.) and confirmation these are enabled via the same single API call without extra per-method integration work.

   • [claimed-docs] “Customers can select the PayPal button on your site and pay with any funding source linked to their PayPal account.”
   • [claimed-docs] “Pay in 4 interest-free payments. Paid every two weeks. No late fees. Won't affect your credit score to apply.”
   • [claimed-docs] “Expanded Checkout includes all offerings from PayPal Checkout and card field customization with your business branding.”
   • [claimed-docs] “Offer up to 400M users1 the choice to pay with PayPal, Venmo, Pay Later, debit and credit cards, and more.”
   • [claimed-docs] “Use your phone and our PayPal app to take contactless cards and digital wallet payments including Apple Pay®, Google Pay, and Samsung Pay.”
   • [claimed-docs] “Accept PayPal payments on your website. Customers can select the PayPal button on your site and pay with any funding source linked to their …”
   • [claimed-docs] “PayPal Open helps you accept payments online, in person, and on the go. Wherever you do business.”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn is a billing/entitlement/metering layer built on top of Stripe (confirmed by autumn-comm-4, which notes it lacks support for other payment providers like Paystack), but the evidence pack contains no mention of Autumn configuring or exposing digital wallets, local payment methods, bank redirects, or BNPL options in checkout. Missing for 10: any documentation of wallet/BNPL/local payment method support, checkout configuration options, or evidence Autumn surfaces Stripe's payment-method capabilities to founders.

   • [community] “Concern raised that Autumn is only built on Stripe, with no support for other payment providers like Paystack, which is needed for regions l…”

   proof ↗⚑ flag

Integration reliability

1. developerSafely retry any API request using idempotency keys so network failures never create duplicate charges

   weight 2 · round drawn
   PayPal—none0/10

   No evidence pack item mentions idempotency keys, PayPal-Request-Id header, or any mechanism for safely retrying API requests to prevent duplicate charges, despite extensive docs on REST API auth, checkout, webhooks, and MCP tools.

   ⚑ flag

   Autumn—none0/10

   No evidence pack item mentions idempotency keys or request-retry safety mechanisms anywhere in Autumn's docs, changelog, or community discussion; the API examples shown (check, track, attach) don't reference idempotency handling.

   ⚑ flag

2. developerRely on signed webhook deliveries with automatic retries, event logs, and replay to keep my system consistent with payment state

   weight 3 · round to PayPal
   PayPal~partialclaimed3/10

   PayPal documents basic webhooks (HTTPS POSTs to a subscribed endpoint) but the evidence pack contains no mention of signature verification/signing, automatic retry behavior, an event log/dashboard, or a replay/redelivery mechanism. Missing for 10: documented webhook signature verification, retry policy details, event log/dashboard access, and replay/resend capability.

   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [claimed-docs] “The initial setup for receiving webhooks is to subscribe a listening URL for a specific app”

   proof ↗⚑ flag

   Autumn—none0/10

   Docs mention webhooks are fired for plan changes, but there is no evidence of signed payloads, automatic retries, event logs, or replay functionality for webhook deliveries; the evidence pack shows only a single event type description with no delivery guarantees.

   • [claimed-docs] “Fired whenever a customer's plans change — new subscriptions, upgrades, downgrades, etc.”

   proof ↗⚑ flag

Onboarding dx

1. developerIntegrate a working payment flow from the docs alone in under a day, with copy-paste examples in my language

   weight 2 · round drawn
   PayPal~partialclaimed6/10

   PayPal's dev docs cover the core building blocks needed for a checkout integration—OAuth token exchange, the Checkout REST flow, webhooks, and a sandbox test environment—plus a mention of generating client code for multiple languages and per-language SDK codegen (Java example shown). However, the evidence pack doesn't show concrete copy-paste code samples in multiple languages (e.g., Node/PHP/Python snippets) nor any independent confirmation that a developer can go from zero to working payment flow in under a day. missing for 10: multi-language copy-paste code snippets, independent time-to-integrate validation, end-to-end quickstart walkthrough evidence.

   • [claimed-docs] “Exchange your client ID and client secret for an access token. The access token authenticates your app when calling PayPal REST APIs.”
   • [claimed-docs] “Customers can select the PayPal button on your site and pay with any funding source linked to their PayPal account.”
   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”
   • [claimed-docs] “Explore our REST API descriptions, generate code for your API clients, and import OpenAPI documents into compatible third-party tools.”
   • [claimed-docs] “npm run codegen-java -- openapi/checkout_orders_v2.json --artifact-id orders”
   • [claimed-docs] “Add server-side endpoints to handle refunds, voids, delayed captures, and specialized checkout flows.”

   proof ↗⚑ flag

   Autumn~partialcommunity6/10

   Docs provide multiple copy-paste TypeScript/JS snippets (check, track, attach, gating) and a CLI/config flow that suggest a fast path to a working payment flow, and community comments (comm-2, comm-5) reflect developers finding it appealing to integrate quickly. However all examples are JS/TS only (no other languages), there's no explicit hands-on confirmation of a 'under a day' integration timeline, and separate security reports (comm-6/7) show at least one integration built from the docs' approach ended up with an IDOR vulnerability, suggesting the docs alone may not guarantee a fully working/secure flow without extra care. Missing for 10: multi-language code samples, an independent hands-on report confirming rapid end-to-end integration, and evidence the documented flow is secure by default.

   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “You can preview your plans without pushing anything: bunx atmn preview”
   • [claimed-docs] “If you already have products created in the dashboard, run `atmn pull` to pull them into your local config.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”
   • [community] “"I've suffered so much with all the pricing changes I've been experimenting with (early stage solo founder)... I'm going to integrate autumn…”
   • [community] “"love this as someone who's been fixing the same billing bugs over and over and who sometimes finds stripe more complex than it should be." …”
   • [community] “Security concern raised: Autumn's server actions were public, unauthenticated routes, meaning "a simple curl request with a different custom…”

   proof ↗⚑ flag

Sandbox fidelity

1. developerBuild and test the entire payment flow in a sandbox or test mode that mirrors production, with test cards for every scenario including declines and disputes

   weight 3 · round to PayPal
   PayPal~partialclaimed5/10

   PayPal documents a dedicated sandbox that 'simulates the live PayPal production environment' (paypal-docs-11), and the agent-toolkit docs explicitly recommend 'Always use the sandbox environment for initial testing' with a PAYPAL_ENVIRONMENT=SANDBOX flag (paypal-supp-toolkit-frameworks), confirming sandbox/production parity for both REST and MCP flows. However, the evidence pack contains no documentation of specific test card numbers or scenario coverage (e.g., simulating declines, chargebacks, or disputes) within the sandbox — only that a live Disputes API and Resolution Center exist for production dispute handling. Missing for 10: documented test card list/scenarios for declines, dispute simulation tooling in sandbox, and independent developer corroboration of sandbox fidelity.

   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”
   • [claimed-docs] “"PayPal's agent toolkit supports the integration of PayPal APIs into AI agent workflows using Amazon Bedrock, CrewAI, LangChain, Model Conte…”
   • [claimed-docs] “Automated operations: Scale dispute management, reduce manual effort, and save time with Disputes API that works seamlessly with your existi…”
   • [claimed-docs] “Disputes API: A programmatic interface that enables you to automate dispute management actions directly from your systems.”

   proof ↗⚑ flag

   Autumn—none0/10

   No evidence in the pack mentions a sandbox/test mode, test cards, or simulated decline/dispute scenarios; Autumn's docs cover CLI, config, gating, webhooks, and MCP but nothing about a testing environment mirroring production payment flows. Since Autumn sits atop Stripe, this is a fair and applicable axis for a billing product, but it's simply undocumented here.

   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “You can preview your plans without pushing anything: bunx atmn preview”
   • [claimed-docs] “If you already have products created in the dashboard, run `atmn pull` to pull them into your local config.”
   • [claimed-docs] “Attaches a plan to a customer. Handles new subscriptions, upgrades and downgrades.”

   proof ↗⚑ flag

Refunds

1. finance-ops userIssue full or partial refunds from both the dashboard and the API

   weight 3 · round to PayPal
   PayPal~partialclaimed6/10

   Evidence confirms API-side refund capability (create_refund, get_refund tools; 'Add server-side endpoints to handle refunds, voids, delayed captures') and dashboard-side refund ability ('Merchants can also add tracking, mark shipments, and issue refunds directly in the PayPal business dashboard'), covering both API and dashboard channels. However, there is no explicit documentation of partial vs full refund distinction, no dedicated Refunds API reference page, and community evidence shows disputes/chargeback handling can be inconsistent, though not specifically contradicting refund issuance itself. Missing for 10: explicit partial-refund API documentation/parameters, a dedicated Refunds API reference, and independent/hands-on confirmation of dashboard refund workflow.

   • [claimed-docs] “Add server-side endpoints to handle refunds, voids, delayed captures, and specialized checkout flows.”
   • [claimed-docs] “Merchants can also add tracking, mark shipments, and issue refunds directly in the PayPal business dashboard.”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”

   proof ↗⚑ flag

   Autumn—none0/10

   No evidence pack items mention refunds at all—no dashboard refund UI, no refund API endpoint, or partial-refund logic; docs cover checks, tracking, attach/subscriptions, credit systems, and CLI/MCP but never refunds.

   ⚑ flag

Multi currency

1. founderCharge customers in their local currencies and settle to my bank in mine with transparent FX handling

   weight 2 · round to PayPal
   PayPal~partialcommunity5/10

   PayPal supports multi-currency transactions (ISO-4217 currency codes, blocking payments in currencies not held, multi-currency payouts) and can accept payments from customers globally, but the evidence pack provides no explicit documentation of automatic FX conversion mechanics, disclosed FX margin/spread, or a transparent rate-and-fee breakdown at settlement — a specific founder concern given community reports of fund holds and account limitations (paypal-comm-3, paypal-comm-8) that could affect predictable settlement. missing for 10: documented FX conversion rate/margin transparency, explicit multi-currency settlement-to-single-bank-account flow, and independent confirmation that FX handling is clear/predictable in practice.

   • [claimed-docs] “To specify currencies in request URI and body parameters, use three-character ISO-4217 codes.”
   • [claimed-docs] “Block or unblock payments sent to you in a currency you don't hold.”
   • [claimed-docs] “Send money to thousands of recipients at once for vendors and disbursements in various currencies.”
   • [claimed-docs] “Send money to thousands of recipients at once for vendors and disbursements in various currencies... A basic solution for sending batch paym…”
   • [community] “I advise people not to depend solely on PayPal because of their tendency to freeze funds over nothing.”
   • [community] “After two eBay/PayPal sales cycles, PayPal emailed: 'money from payments you receive will be placed in a pending balance for up to 21 days..…”

   proof ↗⚑ flag

   Autumn—none0/10

   No evidence in the pack addresses multi-currency charging, FX conversion, or settlement to a founder's home currency; documentation covers pricing models, metering, and Stripe catalog sync only. Community threads even highlight Autumn's dependency on Stripe alone with no other payment processors, but nothing on currency/FX handling either way.

   • [claimed-docs] “Catalog plans now accept a `processors` object so Autumn can reuse your existing Stripe catalog instead of minting new objects.”
   • [community] “Concern raised that Autumn is only built on Stripe, with no support for other payment providers like Paystack, which is needed for regions l…”

   proof ↗⚑ flag

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

   weight 2 · round drawn
   PayPal~partialclaimed6/10

   PayPal documents strong API/UI parity for core merchant flows (checkout, subscriptions, invoicing, payouts, disputes via API vs. Resolution Center, webhooks, platform onboarding) and even an MCP/agent-toolkit layer for natural-language access to many of these (create_order, create_invoice, list_disputes, get_merchant_insights, etc.). However, several consumer-facing PayPal.com UI features (Pay in 4, Savings account APY, debit card cashback, 'pool money' group gifts, Fastlane autofill) have no documented API equivalent, and agentic commerce tools are explicitly gated/limited (e.g., commerce tools 'for gift cards only for now', Store Sync restricted to US physical goods in USD). missing for 10: API parity for consumer money-app features (Pay in 4, Savings, cashback, group pooling), unrestricted access to agentic commerce tools without approval gating.

   • [claimed-docs] “No-code solution: Simplify dispute handling with Resolution Center, a ready-to-use web interface that requires no technical integration.”
   • [claimed-docs] “Automated operations: Scale dispute management, reduce manual effort, and save time with Disputes API that works seamlessly with your existi…”
   • [claimed-docs] “Resolution Center: A web-based dashboard where you can view, monitor, and manually respond to disputes.”
   • [claimed-docs] “Disputes API: A programmatic interface that enables you to automate dispute management actions directly from your systems.”
   • [claimed-docs] “Manage Subscriptions on your account dashboard... Recommended if you don't need to integrate Subscriptions into your product UI.”
   • [claimed-docs] “Manage Subscriptions on your account dashboard - Recommended if you don't need to integrate Subscriptions into your product UI.”
   • [claimed-docs] “The documented agent-toolkit tool inventory spans the merchant back office: catalog (`create_product`, `list_product`, `show_product_details…”
   • [claimed-docs] “Agents can pull business analytics through a dedicated tool: `get_merchant_insights` — "Retrieve business intelligence metrics and analytics…”
   • [claimed-docs] “Store Sync "connects your product catalog and commerce API with PayPal's agentic commerce services, enabling AI agents to discover your prod…”
   • [claimed-docs] “"Merchants can use PayPal's agentic commerce services to create AI-powered shopping experiences, so customers can shop using everyday langua…”
   • [claimed-docs] “Get 3.30% APY11 with your PayPal Savings account”
   • [claimed-docs] “Pool money up front for group gifts and more.”
   • [claimed-docs] “Pay in 4 interest-free payments. Paid every two weeks. No late fees. Won't affect your credit score to apply.”
   • [claimed-docs] “Automatically recognize guest shoppers with Fastlane by PayPal. Next time they shop, we’ll autofill their checkout credentials.”

   proof ↗⚑ flag

   Autumn~partialprobed6/10

   Autumn exposes a broad API (check, track, attach, credit systems, usage-based pricing) plus a CLI (`atmn push/pull/preview`) that can define and sync pricing config as code, and an MCP server for AI-native access to billing/customer/plan tools — strong evidence of API-driven control. However, the docs also show a 'pull' workflow for dashboard-created products, implying some setup still originates in the dashboard UI, and there's no explicit statement or independent test confirming full 1:1 parity between every UI action and API/CLI capability. Missing for 10: explicit parity documentation or proof that all dashboard-only actions (e.g., initial product/catalog creation, Stripe processor linking) are fully replicable via API/CLI without ever touching the UI, and independent verification of this parity.

   • [claimed-docs] “const response = await autumn.check({ customerId: "user_123", featureId: "messages" });”
   • [claimed-docs] “await autumn.track({ customerId: "user_or_org_id_from_auth", featureId: "messages", value: 1, });”
   • [claimed-docs] “Credit systems let you track actions with different credit costs from a single balance pool.”
   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “You can preview your plans without pushing anything: bunx atmn preview”
   • [claimed-docs] “Autumn MCP connects AI assistants to Autumn's billing, customer, plan, balance, and log tools.”
   • [claimed-docs] “If you already have products created in the dashboard, run `atmn pull` to pull them into your local config.”
   • [probe] “official MCP server documented at https://docs.useautumn.com/documentation/mcp”
   • [probe] “official CLI documented at https://docs.useautumn.com/cli/getting-started”

   proof ↗⚑ flag

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

   weight 3 · round drawn
   PayPal—none0/10

   No evidence of any user-facing data export feature or open-format data portability tool; PayPal's evidence pack covers APIs, payments, and AI/agent tooling but nothing about exporting a user's own account/transaction data in open formats or account closure/data portability guarantees.

   ⚑ flag

   Autumn—none0/10

   Autumn's docs show CLI sync/pull for pricing config (autumn.config.ts) and webhooks for event data, but there is no documented feature for exporting all customer/usage/billing data in an open format to allow a user or team to leave the platform. missing for 10: a documented full data-export or account-portability feature, open-format bulk export of customer/usage records, evidence of data portability upon offboarding.

   • [claimed-docs] “The `atmn` CLI lets you define your pricing plans in code via an `autumn.config.ts` file, and sync them to Autumn with a single command.”
   • [claimed-docs] “If you already have products created in the dashboard, run `atmn pull` to pull them into your local config.”
   • [claimed-docs] “Fired whenever a customer's plans change — new subscriptions, upgrades, downgrades, etc.”

   proof ↗⚑ flag

Marketplace

1. developerBuild a marketplace or platform that onboards sellers, splits payments between parties, and pays out to each seller

   weight 2 · round to PayPal
   PayPal✓fullclaimed8/10

   PayPal's Complete Payments Platform explicitly targets marketplaces/platforms with seller onboarding under local regulations, expanded checkout, and multiparty payment processing (paypal-docs-8, paypal-docs-42, paypal-docs-27), while Payouts supports splitting/sending funds to thousands of recipients via email/phone/PayPal ID in various currencies (paypal-docs-5, paypal-docs-24, paypal-docs-41), and REST auth/webhooks/sandbox round out the developer integration story (paypal-docs-1, paypal-docs-9, paypal-docs-11). missing for 10: no explicit hands-on/independent case study of a full marketplace build (onboarding+split+payout) working end-to-end, and details on partner/seller onboarding flows (e.g., Partner Referrals API specifics) are only briefly summarized rather than deeply documented.

   • [claimed-docs] “you can onboard global sellers by applying local regulations and compliance rules”
   • [claimed-docs] “PayPal Complete Payments Platform offers multiparty payment processing solutions for marketplaces and platforms. With these solutions, you c…”
   • [claimed-docs] “Expanded Checkout includes all offerings from PayPal Checkout and card field customization with your business branding.”
   • [claimed-docs] “Send money immediately to multiple recipients with their emails, phone numbers, or PayPal IDs.”
   • [claimed-docs] “Send money to thousands of recipients at once for vendors and disbursements in various currencies.”
   • [claimed-docs] “Send money to thousands of recipients at once for vendors and disbursements in various currencies... A basic solution for sending batch paym…”
   • [claimed-docs] “Exchange your client ID and client secret for an access token. The access token authenticates your app when calling PayPal REST APIs.”
   • [claimed-docs] “Webhooks are HTTPS posts from PayPal to an endpoint on your server whenever the corresponding event type occurs”
   • [claimed-docs] “The PayPal sandbox is a self-contained, virtual testing environment that simulates the live PayPal production environment.”

   proof ↗⚑ flag

   Autumn—none0/10

   Autumn's evidence focuses entirely on single-party SaaS billing—metering, plans, entitlements, customer subscriptions via Stripe—with no mention of multi-party marketplace features like seller onboarding, payment splitting, or payouts (e.g., Stripe Connect-style functionality). No evidence pack item addresses marketplace/platform payment distribution.

   ⚑ flag

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

   weight 2 · round drawn
   PayPal—none0/10

   No evidence in the pack discusses data residency, regional data storage options, or configurable data location for PayPal accounts or its AI tools; this is a fair privacy-posture question for a payments platform handling personal/financial data, but no supporting documentation is present.

   ⚑ flag

   Autumn—none0/10

   Autumn is a billing/entitlements layer for AI and SaaS products; there is no mention anywhere in the evidence pack of data residency, region selection, or data storage location controls. This is an applicable axis for a data-handling SaaS product, but no evidence supports it.

   ⚑ flag

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

   weight 2 · round drawn
   PayPal—none0/10

   No evidence of any telemetry opt-out or usage tracking controls for AI-native users; PayPal's docs focus on payments APIs, and community items discuss account freezes/disputes, not telemetry settings.

   ⚑ flag

   Autumn—none0/10

   Autumn is a billing/entitlements infrastructure product, and its own telemetry/tracking APIs (track, check) are core product features for customer usage billing, not analytics on the developer's own usage of Autumn itself—so the axis of opting out of Autumn's own telemetry collection about its users is applicable but no evidence addresses it. No docs, changelog, or community evidence mention any telemetry opt-out, privacy settings, or data collection disclosure for Autumn's own tooling (CLI, SDK, MCP). missing for 10: any privacy policy or opt-out mechanism documentation, any mention of what usage data Autumn collects from its own users, any CLI/SDK flag to disable telemetry.

   ⚑ flag