Unreal Engine vs PlayCanvas

Agent access

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

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   A direct probe found no llms.txt on Epic's domains, and appending .md to docs URLs returns the HTML SPA shell rather than agent-readable markdown, meaning there is no mechanism to point an agent at machine-readable docs.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed9/10

   A working llms.txt confirmed via HTTP 200 probe plus an unusually large llms-full.txt (2MB) of complete docs is directly served at developer.playcanvas.com, satisfying the exact story of pointing an agent at llms.txt/agent-oriented docs. This is corroborated by both docs claims and independent runtime probes, not just vendor assertion. Missing for 10: no independent third-party (non-probe) report of an agent actually consuming the file successfully in a real workflow.

   • [probe] “PROBE llms.txt: HTTP 200 at https://developer.playcanvas.com/llms.txt # PlayCanvas Developer Documentation > PlayCanvas is an open-source W…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

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

   weight 2 · round to Unreal Engine
   Unreal Engine~partialprobed6/10

   Docs show real automation building blocks: command-line launch arguments (unreal-docs-7, unreal-docs-14), UAT commands for cooking/packaging/deploying (unreal-docs-16, unreal-docs-17), and Python editor scripting for automating asset tasks (unreal-docs-4), all of which are commonly used to run Unreal headlessly in CI pipelines. However, there is no explicit first-party 'headless mode' or CI-integration guide, and the probe notes Unreal's docs/repos are largely non-machine-readable (403s, no llms.txt, no official MCP), suggesting automation setup requires manual/expert configuration rather than a turnkey CI story. Missing for 10: dedicated headless-mode documentation, official CI/CD templates (e.g., GitHub Actions), and independent hands-on confirmation of CI pipelines running Unreal successfully.

   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [claimed-docs] “Command-line Arguments, also called Additional Launch Parameters, customize how the engine runs on startup.”
   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   Evidence shows the engine is npm-installable and there's an Editor MCP server, but the MCP server explicitly requires 'an open PlayCanvas Editor session' (interactive), and no evidence describes headless rendering, CI pipelines, or server-side/automated execution of the engine or editor without a live browser/editor session.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [github] “npm install playcanvas”
   • [community] “When I was working at a mobile game studio, we looked at this as an engine for a project. At the time we decided not to go with it since all…”

   proof ↗⚑ flag

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

   weight 3 · round drawn
   Unreal Engine—none0/10

   Unreal Engine is a platform/tool (not an agent), so the MCP-server axis fairly applies, but evidence shows no official Epic-maintained MCP server exists (GitHub search for `mcp org:EpicGames` returns 0 results) and the most popular community MCP integration is unmaintained since 2025, meaning there's no first-party way to plug MCP servers into Unreal for tool use.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   Evidence only shows PlayCanvas exposing its own Editor as an MCP *server* that AI assistants connect to (playcanvas-docs-6, playcanvas-probe-rt-1) — the opposite direction of this story, which asks whether PlayCanvas itself can plug in external MCP servers to consume their tools. No evidence shows PlayCanvas acting as an MCP client or integrating third-party MCP tool servers into its engine, editor, or workflow.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”

   proof ↗⚑ flag

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

   weight 3 · round to PlayCanvas
   Unreal Engine—none0/10

   Unreal Engine is a platform/tool (not itself an agent), so an official MCP server axis applies, but evidence shows no official Epic MCP server exists — a GitHub org search returns 0 results, and the most popular community MCP integration is unmaintained since 2025.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed9/10

   PlayCanvas documents and ships an official, first-party MCP server (`@playcanvas/editor-mcp-server`, npm-published, official org repo) that connects AI assistants to a live Editor session to modify project data and verify results via the viewport, confirmed both in docs and runtime probes. Missing for 10: independent third-party hands-on validation of the MCP server beyond vendor docs/probe.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”

   proof ↗⚑ flag

5. ai-native userUse an official CLI

   weight 2 · round to Unreal Engine
   Unreal Engine✓fullprobed7/10

   Unreal ships an official CLI surface: UnrealEditor.exe/UAT commands for cooking, packaging, deploying and running projects, plus documented command-line arguments for automation and testing, and Python editor scripting for scriptable workflows — all first-party and independently confirmed as keylessly fetchable per the probe. missing for 10: explicit AI-native framing or agent-oriented tooling around the CLI, and independent hands-on validation of using it in agentic workflows.

   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [claimed-docs] “Command-line Arguments, also called Additional Launch Parameters, customize how the engine runs on startup.”
   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialclaimed4/10

   PlayCanvas ships an official `create-playcanvas` CLI for scaffolding Vite/TypeScript projects, which an AI agent could invoke, but the docs describe it purely as a project scaffolder with no explicit AI-native/agentic features (non-interactive flags, scripting API, automation hooks) and no independent confirmation of agent usage. Missing for 10: documentation of CLI flags/JSON output for automation, evidence of AI-agent usage, and any tie-in with the MCP server workflow.

   • [claimed-docs] “Use `create-playcanvas` to scaffold a Vite and TypeScript project for the Engine, React, or Web Components from one of 12 runnable starters.”

   proof ↗⚑ flag

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

   weight 3 · round to PlayCanvas
   Unreal Engine~partialprobed6/10

   Unreal ships genuinely documented public APIs — a Python editor-scripting API for automating asset/editor tasks, a C++ gameplay API, and command-line/UAT automation flags — which a user or script can drive programmatically. However, the probe evidence shows these are not AI-native accessible: no official Epic MCP server, no llms.txt, docs served only as an HTML SPA (no .md fetch), and the GitHub source repo is EULA/account-gated returning 404 to unauthenticated API calls, with the leading community MCP abandoned since April 2025. Missing for 10: AI-agent-discoverable/machine-readable API docs, an official MCP or agent-facing interface, and keyless programmatic access to source/API metadata.

   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “there may be cases where you want precise control over when and how Assets are discovered, loaded, and audited. For these cases, the Asset M…”
   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [claimed-docs] “Command-line Arguments, also called Additional Launch Parameters, customize how the engine runs on startup.”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed7/10

   PlayCanvas exposes a documented scriptable Editor API for automating/customizing the Editor environment (api.playcanvas.com/editor/) plus an official, npm-published Editor MCP Server explicitly built for AI assistants to modify project data and verify results, alongside a standalone Engine API and llms.txt docs for AI consumption. This gives AI-native users multiple documented, programmatic ways to drive the product. Missing for 10: a formal OpenAPI/REST spec (probe found only 404s for openapi/swagger endpoints) and independent third-party confirmation of API-driven automation success.

   • [claimed-docs] “This module is essential for developers looking to streamline their workflow, create custom editing tools, or integrate external data and sy…”
   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [claimed-docs] “Build applications against the Engine directly. Learn how to use the Engine standalone.”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”
   • [probe] “PROBE openapi: all candidate paths 404 (https://developer.playcanvas.com/openapi.json, https://developer.playcanvas.com/swagger.json, https:…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

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

   weight 2 · round drawn
   Unreal Engine—none0/10

   No evidence anywhere in the pack describes an API credential system, scoped tokens, or least-privilege access controls for AI agents interacting with Unreal Engine or its services; the docs cover Blueprints, C++, plugins, and CLI tools but not credential/permission management. Missing for 10: any mention of API keys/tokens, scoped or role-based access, or agent-specific credential issuance.

   ⚑ flag

   PlayCanvas—none0/10

   Evidence shows an official Editor MCP server connecting AI assistants to an open editor session, but nothing about issuing scoped or least-privilege API credentials/tokens for agent access; no auth/permissions model is documented. Missing for 10: any documentation of API key scoping, permission levels, or credential management for agent access.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”

   proof ↗⚑ flag

8. ai-native userBuild against official SDKs

   weight 2 · round to PlayCanvas
   Unreal Engine~partialprobed5/10

   Unreal ships extensive official build surfaces — C++ APIs, Python editor-scripting API, Blueprint visual scripting, plugin SDK, and CLI/automation tooling — all documented on dev.epicgames.com, which functions as an official SDK for building games/tools. However, the probe evidence shows these are poorly suited for AI-native/agentic consumption: no official MCP server exists, GitHub source access is gated (404 without auth), main sites 403 non-browser clients, no llms.txt, and docs aren't served as fetchable markdown, undercutting AI-native usability of the SDK. Missing for 10: official MCP server or agent-friendly API, machine-readable/llms.txt documentation, keyless programmatic access to source and APIs.

   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “Plugins are collections of code and data that developers can easily enable or disable within the Editor on a per-project basis.”
   • [claimed-docs] “Access to the source code All platforms & features Forums and documentation on the Epic Developer Community”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed8/10

   PlayCanvas publishes multiple official SDKs — the core Engine (npm `playcanvas`), React and Web Components bindings, a `create-playcanvas` scaffolding CLI, a scriptable Editor API, and an official npm-published Editor MCP server for AI automation — all documented and runtime-verified. Community discussion confirms it integrates with standard JS/React workflows, though it also notes some dev-experience friction unrelated to SDK availability. Missing for 10: independent hands-on developer reports specifically validating SDK usage/quality, and a public OpenAPI/versioned reference (probe found only 404s for openapi.json).

   • [claimed-docs] “Use `create-playcanvas` to scaffold a Vite and TypeScript project for the Engine, React, or Web Components from one of 12 runnable starters.”
   • [claimed-docs] “Build applications against the Engine directly. Learn how to use the Engine standalone.”
   • [claimed-docs] “This module is essential for developers looking to streamline their workflow, create custom editing tools, or integrate external data and sy…”
   • [claimed-docs] “Already use React, want declarative 3D scenes, or need to integrate 3D into React apps”
   • [github] “npm install playcanvas”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”
   • [community] “You can use it in conjunction with pretty much any other JS framework available.”

   proof ↗⚑ flag

Agentic features

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

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   No evidence in the pack of Unreal Engine surfacing AI-generated insights or suggestions from project data; docs cover Blueprints, C++, Python scripting, plugins, and networking but nothing about analytics/AI-driven suggestions, and the probe notes the engine has no official MCP or AI-native tooling exposure.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialprobed4/10

   PlayCanvas ships an official Editor MCP server that lets connected AI assistants read and modify project data and verify results in the viewport, which could be used to generate suggestions/insights from a user's scene data, but there is no evidence of a native, built-in AI insights/recommendations feature (e.g., analytics, optimization suggestions) inside the product itself. Missing for 10: dedicated AI-insights UI or documented use-case of the MCP server producing analytical suggestions, independent corroboration of this workflow in practice.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”

   proof ↗⚑ flag

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

   weight 2 · round to Unreal Engine
   Unreal Engine~partialprobed5/10

   Unreal exposes Python editor scripting, commandlets, and UAT/command-line automation that can be scripted to run unattended (e.g., asset processing, cooking, packaging) which supports background automation, but nothing in the evidence describes an AI-native agent framework, scheduler, or official API/MCP for autonomous orchestration — and the probe explicitly notes no official Epic MCP and no machine-readable API, undermining AI-native automation setup. missing for 10: official AI-agent/automation orchestration layer, documented scheduling/trigger system for unattended runs, first-party API or MCP for AI agents to invoke automations.

   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [claimed-docs] “Command-line Arguments, also called Additional Launch Parameters, customize how the engine runs on startup.”
   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialprobed4/10

   PlayCanvas ships an official Editor MCP server that lets AI assistants programmatically modify project data, but it explicitly requires 'an open PlayCanvas Editor session' rather than running as an independent, scheduled, or background process — this is closer to AI-assisted live editing than autonomous background automation. Missing for 10: evidence of scheduled/triggered jobs, headless/background execution without a live editor session, or any autonomous agent loop running unattended.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”

   proof ↗⚑ flag

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

   weight 3 · round to PlayCanvas
   Unreal Engine—none0/10

   No evidence in the pack of any built-in AI assistant inside Unreal Editor that users can delegate tasks to; all documentation covers Blueprints, C++, Python scripting, plugins, and CLI tools, not an AI agent. The probe evidence even notes the lack of any official Epic AI/MCP integration.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “Plugins can add runtime gameplay functionality, modify built-in Engine features (or add new ones), create new file types, and extend the cap…”

   proof ↗⚑ flag

   PlayCanvas~partialprobed5/10

   PlayCanvas does not ship a built-in AI assistant embedded in its UI; instead it exposes an official Editor MCP server that lets external AI assistants (e.g. Claude) connect to an open Editor session and modify/inspect project data via the viewport, effectively allowing task delegation to AI, but the assistant itself lives outside the product rather than inside it. Missing for 10: a first-party, in-product AI assistant UI/chat surface, and any hands-on/community verification that delegation via MCP works well in practice.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”

   proof ↗⚑ flag

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

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   Evidence shows only traditional scripting/programming interfaces (Blueprints, C++, Python, CLI args) with no natural-language command layer; the probe explicitly notes no official Epic MCP server, no llms.txt, and only an unmaintained community MCP project, meaning there's no documented way to operate Unreal via natural-language commands.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”

   proof ↗⚑ flag

   PlayCanvas~partialprobed6/10

   PlayCanvas ships an official, npm-published Editor MCP Server that lets AI assistants modify project data and verify results via the viewport, which is a genuine natural-language/agentic control surface for the Editor. However, this only covers Editor operations, not the runtime Engine or broader workflows, and there's no independent/hands-on evidence of it being used successfully in practice. missing for 10: community/hands-on validation of the MCP server in real workflows, evidence of natural-language control over the Engine/runtime (not just Editor), and broader agentic tool coverage beyond the single MCP server.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”

   proof ↗⚑ flag

Api quality

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

   weight 2 · round drawn
   Unreal Engine—none0/10

   The evidence shows only standard static documentation pages (C++, Blueprints, Python scripting, plugins) with no interactive API reference or runnable/executable code examples; the probe evidence explicitly notes Unreal's docs are not even machine-readable (no llms.txt, .md URLs return HTML shell) and there's no official API sandbox or live-code explorer mentioned anywhere.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   Evidence shows PlayCanvas has API docs (api.playcanvas.com/editor), llms.txt full docs, and starter templates, but no evidence of an interactive API reference page with embedded runnable code examples akin to a live sandbox integrated into the reference itself.

   ⚑ flag

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

   weight 2 · round drawn
   Unreal Engine—none0/10

   No evidence of an OpenAPI or similar machine-readable API spec; the probe evidence explicitly finds Unreal's docs and APIs are not machine-readable (no llms.txt, .md docs serve HTML shells, no official MCP/API spec), confirming absence rather than just lack of mention.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   Explicit probe evidence shows no OpenAPI/Swagger spec exists at any expected path (all candidates 404); while llms.txt and an Editor API doc site exist, there is no machine-readable API spec download offered.

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

   proof ↗⚑ flag

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

   weight 1 · round to PlayCanvas
   Unreal Engine—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

   PlayCanvas~partialclaimed3/10

   PlayCanvas documents Checkpoints for snapshotting/rolling back a project and an MCP server that acts on 'an open PlayCanvas Editor session,' which offers some safety net against irreversible changes, but there is no explicit sandbox/staging environment distinct from production data that an AI agent could test against. missing for 10: dedicated sandbox/staging project mode, explicit production-vs-test data separation, and evidence that MCP or automation actions can't touch the live production project.

   • [claimed-docs] “Checkpoints — Snapshot your project at any point and roll back if needed”
   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [claimed-docs] “With live editing, you can tweak materials, adjust positions, and even modify scripts while your application runs.”

   proof ↗⚑ flag

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

   weight 2 · round drawn
   Unreal Engine—none0/10

   No evidence pack item documents API versioning conventions or a deprecation policy for Unreal's C++/Blueprint/Python APIs; the docs cover feature usage (Blueprints, C++, plugins, replication) but never mention deprecation macros, version compatibility guarantees, or a documented sunset policy. The probe evidence further shows Unreal's docs/API surface is poorly machine-readable, with no llms.txt or markdown API references that would expose such a policy.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   No evidence pack item mentions API versioning semantics or a documented deprecation policy; the OpenAPI probe returned 404s and docs only describe scripting formats, editor features, and an MCP server, not stability guarantees for the API surface.

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

   proof ↗⚑ flag

Agent editor

1. ai-native userAn agent can drive the editor and project programmatically — create scenes and nodes, edit properties, trigger builds — through a documented scriptable interface

   weight 3 · round to PlayCanvas
   Unreal Engine~partialprobed6/10

   Unreal documents a genuine scriptable interface — the Python Editor Scripting API for asset/property automation (docs-4, docs-12), Blueprints for visual node-based logic (docs-1), and CLI/UAT commands for cooking, packaging and triggering builds (docs-7,14,16,17) — which together let a script or agent create/edit content and trigger builds. However, the probe evidence shows no official Epic MCP server or AI-agent-oriented interface exists (0 results for org:EpicGames MCP repos), the most popular community MCP wrapper is unmaintained since 2025, and the docs themselves are not agent-friendly (SPA HTML, no llms.txt, .md suffix doesn't return markdown), undermining ease of programmatic/agentic access despite the underlying scripting capability being real. Missing for 10: official AI-agent/MCP integration, machine-readable documentation, evidence of an agent actually driving these interfaces end-to-end.

   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “there may be cases where you want precise control over when and how Assets are discovered, loaded, and audited. For these cases, the Asset M…”
   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed8/10

   PlayCanvas ships an official, npm-published Editor MCP server that lets AI agents connect to an open Editor session, modify project data, and verify results, plus a separate documented scriptable Editor API (api.playcanvas.com/editor) for programmatic scene/node/property manipulation and automation. This directly satisfies the story of an agent driving the editor programmatically through a documented interface. Missing for 10: no independent/hands-on community report of an agent actually using the MCP server end-to-end, and no explicit documentation of triggering builds via this interface.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [claimed-docs] “This module is essential for developers looking to streamline their workflow, create custom editing tools, or integrate external data and sy…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

Codegen

1. ai-native userLLM-generated code mostly works on the first try — stable, well-documented APIs with abundant public examples the models have seen

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   No evidence directly addresses whether LLM-generated Unreal C++/Blueprint code works reliably on first try; the pack only shows general API docs exist and community comments about C++ complexity and blueprint limitations. The probe evidence even indicates docs are poorly machine-readable (403s, no llms.txt, gated source) which undercuts the 'abundant examples models have seen' premise but isn't a direct test of code-gen success either.

   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [community] “My main problem with UE is C++, I don't want to even try to download files or read json in a blueprint.”
   • [community] “blueprints are great for prototyping, but once you have too much complexity in a BP it can easily cascade inefficiencies to the point it bec…”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialprobed6/10

   PlayCanvas offers strong LLM-friendly infrastructure — a dedicated llms.txt/llms-full.txt (over 2MB of docs), a stable, long-standing npm-published engine (2.22.2, 48k weekly downloads) with abundant public code examples and community history, plus scripting API docs — all of which support an LLM having 'seen' the API. However, community threads note real dev-experience quirks and unusual design choices that could trip up naive LLM-generated code, and there's no direct hands-on evidence of first-try LLM code generation success rates. Missing for 10: independent benchmark or testimonial confirming LLM-generated PlayCanvas code works correctly on first try, and no discussion of API stability/breaking-change history that would affect training-data staleness.

   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”
   • [probe] “PROBE llms.txt: HTTP 200 at https://developer.playcanvas.com/llms.txt # PlayCanvas Developer Documentation > PlayCanvas is an open-source W…”
   • [github] “npm install playcanvas”
   • [community] “That issue was opened in 2022. Honestly if you really get into the nitty gritty of PC it becomes quickly apparent there's some very strange …”
   • [community] “I think Playcanvas would do well to focus on the basics, like being able to export/import variables in the standard js way, being able to de…”

   proof ↗⚑ flag

Copilot

1. ai-native userAn official AI assistant inside the engine helps with engine tasks — generating scripts, answering API questions, scaffolding scenes or assets

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   No evidence of an official in-engine AI assistant for generating scripts, answering API questions, or scaffolding scenes; the pack only covers Blueprints, C++, Python editor scripting, and plugins as manual tooling. The probe evidence explicitly notes no official Epic MCP/AI integration and that the most popular community MCP project is unmaintained.

   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”

   proof ↗⚑ flag

   PlayCanvas~partialprobed6/10

   PlayCanvas ships an official, npm-published Editor MCP Server that lets external AI assistants connect to a live Editor session to modify project data, script entities, and verify changes via the viewport—directly enabling AI-assisted script generation and scene/asset scaffolding (playcanvas-docs-6, playcanvas-probe-rt-1). However, this is an MCP bridge for third-party assistants rather than a built-in chat/assistant panel inside the engine UI, and there's no dedicated in-engine API Q&A assistant beyond the llms.txt docs feed. Missing for 10: a native in-editor chat/assistant UI, and explicit evidence of it answering free-form API questions rather than just executing structured edits.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

Formats

1. technical artistStandard interchange formats — glTF, FBX, USD — import cleanly

   weight 2 · round drawn
   Unreal Engine—none0/10

   The evidence pack contains no documentation, community, or probe citations addressing glTF, FBX, or USD import functionality in Unreal Engine; all provided docs cover Blueprints, C++, plugins, networking, and licensing instead. Since asset-pipeline import is a fair and applicable axis for a game engine, but no supporting evidence exists in this pack, the verdict is none.

   ⚑ flag

   PlayCanvas—none0/10

   The evidence pack contains no mention of glTF, FBX, or USD import support, nor any asset-pipeline/import documentation; it covers editor, scripting, MCP server, and community sentiment unrelated to file format import.

   ⚑ flag

Import

1. technical artistAsset import is automatable — import hooks, presets, and pipeline scripts process incoming assets without hand-clicking each one

   weight 2 · round to Unreal Engine
   Unreal Engine~partialclaimed6/10

   Docs confirm Python scripting can automate asset-management tasks like LOD generation, and the Asset Manager gives programmatic control over asset discovery/loading, while Plugins can add new file-type handling — together these support scripted, non-interactive asset pipelines. However, there's no explicit documentation of dedicated 'import hooks' (e.g. post-import delegates/factories) or preset-based automated import workflows, and no independent/community confirmation that artists actually use this for hands-off batch import. missing for 10: explicit import-hook/factory API docs, named import-preset tooling, and third-party/community validation of automated import pipelines.

   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “provides a framework for managing Assets that can divide content into chunks that make sense in the context of your project”
   • [claimed-docs] “there may be cases where you want precise control over when and how Assets are discovered, loaded, and audited. For these cases, the Asset M…”
   • [claimed-docs] “Plugins can add runtime gameplay functionality, modify built-in Engine features (or add new ones), create new file types, and extend the cap…”
   • [claimed-docs] “Plugins are collections of code and data that developers can easily enable or disable within the Editor on a per-project basis.”

   proof ↗⚑ flag

   PlayCanvas~partialcommunity3/10

   The Editor API is explicitly billed for scripting custom tools and integrating external systems (playcanvas-docs-9), and the community references playcanvas-sync for mapping project assets to a local filesystem (playcanvas-comm-10) and the MCP server for programmatic project data changes (playcanvas-docs-6), all of which imply some automation capability. However, there is no documented import-hook system, asset presets, or a dedicated pipeline-scripting API specifically for processing incoming assets automatically. Missing for 10: explicit asset import hooks, configurable import presets, documented pipeline scripts triggered on asset ingestion, and independent confirmation these tools eliminate manual per-asset clicking.

   • [claimed-docs] “This module is essential for developers looking to streamline their workflow, create custom editing tools, or integrate external data and sy…”
   • [community] “I recommend checking out the newly released playcanvas-sync tools which let you map your PlayCanvas project's assets to your local machine's…”
   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”

   proof ↗⚑ flag

Marketplace

1. game developerA large asset store or package ecosystem gives me ready-made models, tools, and plugins

   weight 2 · round to Unreal Engine
   Unreal Engine~partialprobed5/10

   Docs show a genuine plugin and asset-management ecosystem (plugins can add gameplay features, new file types, editor extensions; Asset Manager for content organization) and community comment notes strong third-party program/community support, but there is no direct evidence in the pack describing the actual Marketplace/Fab asset store, its catalog size, or the quality of ready-made models — the probe entry only notes fab.com exists but is API-gated, not its content richness. Missing for 10: concrete marketplace/store catalog evidence, examples of ready-made models/tools available, and independent user reviews of the store experience.

   • [claimed-docs] “Plugins can add runtime gameplay functionality, modify built-in Engine features (or add new ones), create new file types, and extend the cap…”
   • [claimed-docs] “Plugins are collections of code and data that developers can easily enable or disable within the Editor on a per-project basis.”
   • [claimed-docs] “provides a framework for managing Assets that can divide content into chunks that make sense in the context of your project”
   • [claimed-docs] “there may be cases where you want precise control over when and how Assets are discovered, loaded, and audited. For these cases, the Asset M…”
   • [community] “Unreal was already the best engine purely on being easy to work with, support from third party programs and community, none of those things …”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   The evidence pack shows PlayCanvas has an editor, engine, scripting system, MCP server, and npm-published engine, but nowhere mentions an asset store, marketplace, or plugin/package ecosystem for ready-made models or tools. No community or docs citation references anything resembling a store of third-party assets or plugins.

   ⚑ flag

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

   weight 2 · round to Unreal Engine
   Unreal Engine~partialprobed5/10

   Unreal's Python editor-scripting API and Asset Manager are documented to automate time-consuming asset tasks (e.g., batch-generating LODs for Static Meshes) and manage large sets of Assets in chunks, which supports bulk operations across many items, and UAT/command-line tooling and cooking commands support batch build/deploy across multiple platforms/targets. However there is no concrete first-party example of an AI-native agent performing bulk edits, no official MCP server, and the probe evidence flags Unreal as the least machine-readable/agent-accessible of comparable products (gated APIs, 403s, no llms.txt), undercutting AI-native automation-depth claims. Missing for 10: worked examples of bulk operations orchestrated end-to-end, official AI/agent tooling or MCP integration, independent hands-on confirmation of scripted bulk workflows.

   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “provides a framework for managing Assets that can divide content into chunks that make sense in the context of your project”
   • [claimed-docs] “there may be cases where you want precise control over when and how Assets are discovered, loaded, and audited. For these cases, the Asset M…”
   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialprobed4/10

   PlayCanvas exposes a scriptable Editor API (api.playcanvas.com/editor) explicitly for building custom editing tools and integrating external data, and an official MCP server that lets AI assistants modify project data programmatically — both imply automation potential across many assets, but neither doc explicitly describes bulk/batch operations (e.g., multi-entity edits, batch asset import) or gives a hands-on example of many-item automation. Missing for 10: explicit bulk-edit/batch API examples, documented multi-item scripting workflows, and independent confirmation of large-scale automation via the MCP server or Editor API.

   • [claimed-docs] “This module is essential for developers looking to streamline their workflow, create custom editing tools, or integrate external data and sy…”
   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”

   proof ↗⚑ flag

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

   weight 3 · round to Unreal Engine
   Unreal Engine✓fullcommunity6/10

   Unreal's Blueprint visual scripting and C++ delegate system explicitly support binding functions to be invoked automatically at a future time/event (dynamic delegates), and Blueprints expose event graphs for designers to wire up event-triggered logic without code. Community commentary confirms Blueprints are used for prototyping event-driven logic though complexity pushes users to C++. Missing for 10: explicit documentation of Event Dispatchers/Event Graph event-binding UI, and independent hands-on validation of rule-trigger reliability at scale.

   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [claimed-docs] “You can dynamically bind a delegate to a member function of an arbitrary object, calling the function on the object at a future time”
   • [community] “blueprints are great for prototyping, but once you have too much complexity in a BP it can easily cascade inefficiencies to the point it bec…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   The evidence pack covers scripting syntax, editor collaboration, physics components, and the MCP server for AI-assisted project editing, but contains no documentation of an explicit event/rule system where users can define triggers that automatically fire actions (e.g., event listeners, condition-action bindings) for automation purposes. Missing for 10: documentation of an event-trigger/rule engine, examples of defining conditional automation, and evidence tying this to AI-native workflows.

   ⚑ flag

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

   weight 1 · round to PlayCanvas
   Unreal Engine—none0/10

   The evidence pack covers Blueprints, C++, Python scripting, plugins, asset management, and cooking, but contains no mention of version control integration, change review/diffing, or rollback mechanisms for Blueprints, Python scripts, or other automations. This is a plausible axis for a dev engine (e.g., source control on Blueprints), but no supporting evidence exists.

   ⚑ flag

   PlayCanvas~partialcommunity5/10

   PlayCanvas Editor has a Checkpoints feature to snapshot a project and roll back, which covers basic versioning/rollback of editor state (including changes made via the MCP-driven AI automation), but there's no documented diff/review workflow for automations and community feedback explicitly flags the lack of standard source-control integration as a pain point. missing for 10: dedicated review/diff tooling for automation changes, git-style commit history, and any evidence that checkpoints integrate specifically with AI-driven MCP edits.

   • [claimed-docs] “Checkpoints — Snapshot your project at any point and roll back if needed”
   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [community] “I think Playcanvas would do well to focus on the basics, like being able to export/import variables in the standard js way, being able to de…”

   proof ↗⚑ flag

2d

1. game developer2D is a first-class workflow — sprites, tilemaps, 2D physics — not a 3D afterthought

   weight 2 · round drawn
   Unreal Engine—none0/10

   The evidence pack contains no mention of sprites, tilemaps, Paper2D, or 2D physics support in Unreal Engine—only C++, Blueprints, plugins, networking, and licensing topics. Since a 3D-first engine could plausibly ship a first-class 2D workflow, this axis applies, but no evidence documents it, so the story is unsupported.

   ⚑ flag

   PlayCanvas—none0/10

   Evidence pack covers 3D features (rigidbody/collision physics, Gaussian Splatting, Engine, Editor) but contains no mention of sprites, tilemaps, or a 2D physics system — PlayCanvas is documented here purely as a 3D-first engine. This is a fair question for a game engine, but no evidence supports a first-class 2D workflow, so it's 'none' rather than 'na'.

   • [claimed-docs] “The rigidbody component decides how an entity moves, the collision component gives it a physical shape, and the alpha joint component constr…”
   • [github] “Gaussian Splatting - First-class support for loading and rendering 3D Gaussian Splats”

   proof ↗⚑ flag

Performance

1. game developerProfile and scale — a real profiler, plus data-oriented paths (ECS, jobs, instancing) when scenes get heavy

   weight 2 · round drawn
   Unreal Engine—none0/10

   The evidence pack contains no documentation or community evidence about Unreal Insights, stat/profiling commands, GPU profiling, or data-oriented paths like Mass Entity/ECS, job systems, or GPU instancing — only generic command-line-argument docs mentioning 'testing and optimizing' in passing. Profiling and data-oriented scaling are core, expected capabilities for a AAA game engine, so the axis applies, but nothing in the pack demonstrates a real profiler or ECS/jobs/instancing systems.

   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [claimed-docs] “Command-line Arguments, also called Additional Launch Parameters, customize how the engine runs on startup.”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   The evidence pack contains no mention of a profiler, ECS architecture, job/worker system, or GPU instancing for scaling heavy scenes — only physics components, Gaussian splatting, and editor/collab features are documented. This is a fair axis for any game engine, but PlayCanvas's pack simply doesn't address it.

   ⚑ flag

Physics

1. game developerBuilt-in physics — rigid bodies, collisions, raycasts — works out of the box

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   The evidence pack contains no documentation, community reports, or probes specifically addressing Unreal's physics system (rigid bodies, collision detection, raycasting, Chaos/PhysX). Coverage is entirely about Blueprints, C++, plugins, asset management, networking, licensing, and community sentiment on cost/usability — none of which substantiates the built-in physics story.

   ⚑ flag

   PlayCanvas~partialcommunity6/10

   Docs confirm built-in rigidbody, collision, and joint components enabling physics simulation out of the box (playcanvas-docs-8), and community sources corroborate PlayCanvas as a capable, well-designed engine used in shipped games. However, raycasts are not explicitly mentioned, and there's no hands-on evidence or tutorial demonstrating physics working smoothly without setup friction (e.g., Ammo.js/physics engine integration steps). Missing for 10: explicit raycast documentation, independent hands-on confirmation of zero-config physics setup, and detail on which physics backend is bundled vs. requires separate install.

   • [claimed-docs] “The rigidbody component decides how an entity moves, the collision component gives it a physical shape, and the alpha joint component constr…”
   • [community] “THREE.js is very low-level framework, while PlayCanvas additionally to low-level engine access has game framework, which will allow you to c…”
   • [community] “I haven't used it, but I remember looking at the minified code a while back when it was first released. It looks very well done, by someone …”

   proof ↗⚑ flag

Rendering

1. game developerThe engine ships a modern production 3D renderer — PBR materials, global illumination or baked lighting, shadows, post-processing

   weight 3 · round to Unreal Engine
   Unreal Engine~partialcommunity5/10

   Community evidence confirms Unreal's renderer includes Lumen for real-time global illumination/reflections and general praise for lighting quality, supporting the GI/rendering claim, but the evidence pack lacks first-party documentation or hands-on confirmation of PBR material workflows, shadow systems, or the post-processing pipeline explicitly. missing for 10: explicit docs/evidence on PBR material editor, shadow mapping/ray-traced shadows, and post-process volume/effects stack.

   • [community] “Lumen implements efficient Software Ray Tracing, allowing for global illumination and reflections to run on a wide range of video cards, whi…”
   • [community] “For someone who's actually using UE4 as an artistic break for web dev, I thoroughly recommend it. It allows you to create great environments…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   The evidence pack documents scripting, editor collaboration, physics, MCP integration, and Gaussian splatting support, but contains no first-party or community evidence about PBR materials, global illumination/baked lighting, shadow systems, or a post-processing pipeline. Per the rules, absence of evidence for this applicable rendering-feature axis must be scored as 'none' rather than inferred from outside knowledge.

   ⚑ flag

Collaboration

1. studio leadThe project format and tooling play well with version control and multi-person teams — mergeable scenes, diffable text formats, or built-in collaboration

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   The evidence pack contains no mention of Unreal's scene/level file formats, diffability, source-control integration (e.g., Perforce/Git-LFS workflows), or any built-in multi-user collaboration tooling; it only covers Blueprints, C++, plugins, networking replication, and licensing. Missing for 10: any documentation or community evidence on .uasset/.umap format mergeability, text-based diffable formats, or built-in collaboration features like Multi-User Editing.

   ⚑ flag

   PlayCanvas!disputedcontradicted4/10

   PlayCanvas markets real-time multi-user editing as solving merge conflicts entirely (docs-4, docs-13) rather than a diffable/VCS-friendly text format, but hands-on community reports concretely contradict the 'plays well with version control' framing: developers note the project lives on PlayCanvas's servers with 'no easy local dev environment' and that basics like 'using your own source control' are missing, with unofficial third-party tools (playcanvas-sync) needed just to sync assets locally, and uploading being a 'deal breaker' for offline/local workflows. Checkpoints (docs-5) offer some rollback safety but are not the same as git-based diff/merge tooling. missing for 10: official git-friendly/text-diffable scene format, first-party local-file sync equivalent to real VCS integration, and evidence the community-reported local-dev gap has been resolved.

   • [claimed-docs] “Multiple developers, artists, and designers can build the same scene simultaneously—no merge conflicts, no stepping on each other's toes.”
   • [claimed-docs] “With live editing, you can tweak materials, adjust positions, and even modify scripts while your application runs.”
   • [claimed-docs] “Checkpoints — Snapshot your project at any point and roll back if needed”
   • [community] “I think Playcanvas would do well to focus on the basics, like being able to export/import variables in the standard js way, being able to de…”
   • [community] “When I was working at a mobile game studio, we looked at this as an engine for a project. At the time we decided not to go with it since all…”
   • [community] “I think the engine is open source; but the real appeal is using the real time collab online tool, which is proprietary and had the drawbacks…”
   • [community] “I'd love to be able to use their scene editor offline, uploading assets is a deal breaker.”
   • [community] “I recommend checking out the newly released playcanvas-sync tools which let you map your PlayCanvas project's assets to your local machine's…”

   proof ↗⚑ flag

Editor

1. technical artistA full visual editor — viewport, inspector, prefabs/scene composition — is the primary way to build levels

   weight 3 · round to Unreal Engine
   Unreal Engine✓fullcommunity7/10

   Unreal's editor-centric workflow (Blueprints visual scripting, Editor extensions, Asset Manager, Level building/testing via UE Toolbar) is well documented as the primary way to construct levels and gameplay without programmer-only tools, and community comments (comm-3, comm-11, comm-16) corroborate that it's an actual GUI-driven editor experience artists use. missing for 10: explicit documentation naming 'viewport' and 'inspector' panels, and 'prefab' terminology (Unreal uses Blueprint/Actor classes instead) with independent hands-on confirmation of scene-composition workflow specifically for technical artists.

   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [claimed-docs] “Plugins can add runtime gameplay functionality, modify built-in Engine features (or add new ones), create new file types, and extend the cap…”
   • [claimed-docs] “provides a framework for managing Assets that can divide content into chunks that make sense in the context of your project”
   • [claimed-docs] “Use UE Toolbar to quickly test a part of your currently loaded Level for testing and debugging.”
   • [community] “Unreal was already the best engine purely on being easy to work with, support from third party programs and community, none of those things …”
   • [community] “What I hate about unity is the whole interface editor thing... Is this Unreal Engine more earth to earth when it comes to programming?”

   proof ↗⚑ flag

   PlayCanvas~partialcommunity6/10

   Docs confirm a full drag-and-drop visual Editor with real-time collaborative scene building, live editing of materials/positions, and checkpoint/versioning (playcanvas-docs-3, docs-4, docs-5, docs-13), which matches the viewport+scene-composition part of the story. However, no evidence explicitly documents an 'inspector' panel or a 'prefab' system by name, and community reports flag real dev-experience friction (no easy local dev, assets must be uploaded, offline editing not supported) that complicate treating the editor as the primary, frictionless build workflow. missing for 10: explicit inspector/prefab terminology and documentation, independent hands-on confirmation of prefab workflow, resolution of community-reported offline/local-dev friction.

   • [claimed-docs] “Want drag-and-drop scene building, real-time collaboration, or prefer visual tools over code”
   • [claimed-docs] “Multiple developers, artists, and designers can build the same scene simultaneously—no merge conflicts, no stepping on each other's toes.”
   • [claimed-docs] “Checkpoints — Snapshot your project at any point and roll back if needed”
   • [claimed-docs] “With live editing, you can tweak materials, adjust positions, and even modify scripts while your application runs.”
   • [community] “When I was working at a mobile game studio, we looked at this as an engine for a project. At the time we decided not to go with it since all…”
   • [community] “I'd love to be able to use their scene editor offline, uploading assets is a deal breaker.”

   proof ↗⚑ flag

Extensibility

1. game developerExtend and automate the editor itself — custom tools, editor scripts, plugins that manipulate scenes and assets programmatically

   weight 3 · round to Unreal Engine
   Unreal Engine✓fullcommunity8/10

   Docs confirm multiple official extension paths: Python editor scripting for automating asset/editor tasks, Blueprints for designer-level scripting, C++ for deep gameplay/editor class extension, and a Plugins system explicitly designed to 'extend the capabilities of the Editor.' These are core, well-documented first-party mechanisms for programmatic scene/asset manipulation and editor customization. Missing for 10: independent hands-on corroboration specifically of the Python/editor-scripting API (community evidence mostly discusses C++/Blueprint complexity rather than editor automation), and no mention of newer scripting surfaces like the Editor Utility Widgets or Python API depth in third-party sources.

   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “Plugins can add runtime gameplay functionality, modify built-in Engine features (or add new ones), create new file types, and extend the cap…”
   • [claimed-docs] “Plugins are collections of code and data that developers can easily enable or disable within the Editor on a per-project basis.”
   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [community] “blueprints are great for prototyping, but once you have too much complexity in a BP it can easily cascade inefficiencies to the point it bec…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed7/10

   PlayCanvas explicitly documents a scriptable Editor API module (api.playcanvas.com/editor) built 'for developers looking to... create custom editing tools, or integrate external data and systems into the PlayCanvas Editor environment,' plus an official MCP server for AI-driven automation of live Editor sessions, and community-built sync tools for local file mapping — together showing programmatic extension of the editor itself. Missing for 10: concrete example plugins/marketplace or hands-on demonstration of a third-party editor tool built with this API, and no independent review confirming ease of use for this specific API (only general dev-experience complaints exist).

   • [claimed-docs] “This module is essential for developers looking to streamline their workflow, create custom editing tools, or integrate external data and sy…”
   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”
   • [community] “I recommend checking out the newly released playcanvas-sync tools which let you map your PlayCanvas project's assets to your local machine's…”

   proof ↗⚑ flag

Ci

1. game developerBuild and export the project from the command line, headless, in CI — no human clicking an editor

   weight 3 · round to Unreal Engine
   Unreal Engine✓fullprobed8/10

   Official docs give concrete CLI commands for headless building/cooking/packaging/deploying a project (UnrealEditor.exe -run=cook, and the build/cook/package/deploy/run automation command), plus a dedicated command-line-arguments reference for scripted/CI use, all without needing to open the editor UI. Missing for 10: independent third-party CI pipeline case study confirming it works end-to-end, and explicit mention of UAT/BuildCookRun tool name or exit-code/scripting integration details.

   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [claimed-docs] “Command-line Arguments, also called Additional Launch Parameters, customize how the engine runs on startup.”
   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialcommunity4/10

   The standalone Engine is a plain npm package usable with Vite/TypeScript scaffolding, so Engine-only projects can plausibly be built and bundled headlessly via a standard `vite build` in CI (playcanvas-gh-2, playcanvas-docs-1, playcanvas-docs-7). However, PlayCanvas's primary workflow is the cloud Editor, and community reports confirm no easy local dev environment or native CLI export — code/assets live on PlayCanvas's servers, requiring third-party 'playcanvas-sync' tooling and manual tinkering to approximate a local/CI workflow (playcanvas-comm-8, playcanvas-comm-9, playcanvas-comm-10). No first-party CLI/CI build-and-export tool for Editor projects is documented anywhere in the evidence. Missing for 10: an official CLI/CI export command for Editor-based projects, documentation of headless Editor project builds, and independent confirmation that Editor-based CI pipelines work reliably.

   • [github] “npm install playcanvas”
   • [claimed-docs] “Use `create-playcanvas` to scaffold a Vite and TypeScript project for the Engine, React, or Web Components from one of 12 runnable starters.”
   • [claimed-docs] “Build applications against the Engine directly. Learn how to use the Engine standalone.”
   • [community] “When I was working at a mobile game studio, we looked at this as an engine for a project. At the time we decided not to go with it since all…”
   • [community] “It's certainly not super obvious to setup but there are ways to use PlayCanvas with local setup... with some tinkering you can have hot modu…”
   • [community] “I recommend checking out the newly released playcanvas-sync tools which let you map your PlayCanvas project's assets to your local machine's…”

   proof ↗⚑ flag

Server

1. game developerGame logic runs headless on servers — a dedicated-server or server-runtime build without rendering

   weight 2 · round drawn
   Unreal Engine—none0/10

   The evidence pack covers general networking/replication (client-server sync), command-line arguments, and cooking/packaging, but nowhere mentions Unreal's dedicated-server build target, a headless/no-rendering server runtime, or how to produce a server-only executable. Without explicit documentation of a server-only build path, this axis is unsupported by the given evidence. Missing for 10: any mention of a 'Server' build target, -server or -nullrhi headless launch flags, or dedicated-server packaging instructions.

   • [claimed-docs] “Replication is the name for the process of synchronizing data and procedure calls between clients and servers”
   • [claimed-docs] “Replication is the name for the process of synchronizing data and procedure calls between clients and servers.”
   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [claimed-docs] “Command-line Arguments, also called Additional Launch Parameters, customize how the engine runs on startup.”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   The evidence pack shows PlayCanvas as a browser-focused WebGL/WebGPU engine with npm packages, Editor, and MCP server tooling, but nothing about a dedicated headless/server-runtime build without rendering for running game logic on servers. Missing for 10: any docs or GitHub evidence of a headless Node.js server build, server-side simulation mode, or rendering-free runtime.

   • [claimed-docs] “Build applications against the Engine directly. Learn how to use the Engine standalone.”
   • [github] “npm install playcanvas”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

Testing

1. game developerTests run headlessly — unit and integration tests of game code execute in CI against the real engine

   weight 2 · round drawn
   Unreal Engine—none0/10

   The evidence pack only shows generic command-line argument docs and a UE Toolbar mention for ad-hoc level testing/debugging; there is no evidence of Unreal's Automation System, Gauntlet, or any documented headless unit/integration test execution flow running the real engine in CI. Missing for 10: mention of the Automation Testing framework, -nullrhi/-unattended/-ExecCmds test invocation, CI integration examples, or independent reports of headless test runs against the real engine.

   ⚑ flag

   PlayCanvas—none0/10

   No evidence of a testing framework, headless mode, or CI integration for running unit/integration tests against the PlayCanvas engine; docs cover scripting, editor, MCP server, and standalone engine usage but never mention headless test execution or CI workflows.

   ⚑ flag

Governance

1. studio leadThe project's governance and funding are transparent — a foundation, published finances, or a public roadmap I can plan against

   weight 1 · round drawn
   Unreal Engine—none0/10

   Evidence covers licensing costs, royalty terms, and technical documentation, but nothing indicates a governing foundation, published financial statements, or a public roadmap for Unreal Engine — Epic Games remains a private, closed-governance company with only EULA/royalty terms disclosed (unreal-docs-10, unreal-comm-9, unreal-comm-13).

   • [claimed-docs] “Access to the source code”
   • [community] “For those who don't know what the deal is with Unreal Engine, there's no charge until you get the first million dollars in revenue. Then the…”
   • [community] “It appears they've backed off on the royalties, too: originally, it was 5% of gross, and now it's 5% of gross after $3000 sales per quarter.”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   No evidence of a governing foundation, published financials, or a public roadmap; PlayCanvas is run as a commercial product (subscription plans mentioned) with no transparency artifacts cited. Community comments discuss engine features and dev-experience, not governance or funding.

   ⚑ flag

License

1. studio leadThe license is permissive with no royalties or per-install fees — I keep what my game earns

   weight 3 · round to PlayCanvas
   Unreal Engine—none0/10

   Community evidence explicitly documents that Unreal charges a 5% royalty on gross revenue after the first $1M (or $3,000/quarter) — this is the opposite of 'no royalties, keep what you earn.' No evidence pack item supports a royalty-free/no-fee licensing model; instead multiple citations (comm-1, comm-9, comm-13, comm-2) confirm a royalty-based commercial license.

   • [community] “UE offers everything free... Only way to pay to use Unreal Engine is to earn more than $1,000,000 USD which for most people means everything…”
   • [community] “For those who don't know what the deal is with Unreal Engine, there's no charge until you get the first million dollars in revenue. Then the…”
   • [community] “It appears they've backed off on the royalties, too: originally, it was 5% of gross, and now it's 5% of gross after $3000 sales per quarter.”
   • [community] “i know games with $3 million dollar budgets that chose Unity over Unreal because with Unreal that would have cost them $150k whereas with Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed7/10

   The core PlayCanvas Engine is confirmed MIT-licensed and open source ('MIT engine runtime' in playcanvas-probe-rt-2; community confirms 'the engine is open source' in playcanvas-comm-11 and playcanvas-comm-16), meaning a studio can ship and monetize games without royalties or per-install fees on the runtime itself. However, evidence also shows the hosted Editor/collaboration tooling is a paid subscription (e.g., $15/month for private projects, playcanvas-comm-4), which is a studio-tooling cost rather than a revenue royalty, so it doesn't contradict the engine's permissive licensing but adds nuance. Missing for 10: explicit vendor documentation stating 'no royalties/no revenue share' policy and independent legal/community confirmation of unrestricted commercial use terms.

   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”
   • [community] “I think the engine is open source; but the real appeal is using the real time collab online tool, which is proprietary and had the drawbacks…”
   • [community] “It's an open source javascript game engine that is kinda like halfway between something like three.js and e.g. Unity—pretty lightweight, but…”
   • [community] “Every time I've taken a look at this engine it has always felt very limited. And with a $15/month subscription just to be able to create pri…”

   proof ↗⚑ flag

Source

1. game developerRead and modify the full engine source when I hit a wall

   weight 2 · round to PlayCanvas
   Unreal Engine✓fullprobed7/10

   Epic's license page explicitly grants 'Access to the source code' and docs describe creating new Gameplay C++ classes that recompile directly into the Editor, and community comments corroborate 'source code included.' Missing for 10: independent hands-on account of actually patching core engine internals (not just gameplay classes), and the probe notes the source repo is EULA/GitHub-account gated rather than freely open, which adds friction to the 'when I hit a wall' scenario.

   • [claimed-docs] “Access to the source code”
   • [claimed-docs] “Access to the source code All platforms & features Forums and documentation on the Epic Developer Community”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [community] “Crytek has financial worries. Plus it's the most difficult to use, because it's almost all C++. Unity is very accessible thanks to C# script…”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed8/10

   The PlayCanvas Engine is confirmed open source (MIT) and published on GitHub/npm, allowing developers to read and modify the core engine code directly, corroborated by community comment that 'the engine is open source' and runtime probes confirming the MIT license and public repo. Missing for 10: explicit contribution/build docs or a first-party statement of the license terms, and no independent hands-on account of someone actually patching engine internals.

   • [github] “Gaussian Splatting - First-class support for loading and rendering 3D Gaussian Splats”
   • [github] “npm install playcanvas”
   • [community] “I think the engine is open source; but the real appeal is using the real time collab online tool, which is proprietary and had the drawbacks…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

Trust

1. studio leadThe pricing and license terms have a track record of stability — no retroactive changes that reprice games already shipped

   weight 2 · round drawn
   Unreal Engine—none0/10

   The pack shows royalty terms changed over time (5% of gross → 5% after $3,000/quarter) but no evidence addresses whether such changes were ever applied retroactively to already-shipped titles or whether Epic has an explicit no-retroactive-repricing policy/track record. Missing for 10: documented policy or community confirmation that license terms are grandfathered for shipped games, evidence of any past retroactive-change controversy or its resolution.

   • [community] “It appears they've backed off on the royalties, too: originally, it was 5% of gross, and now it's 5% of gross after $3000 sales per quarter.”
   • [community] “For those who don't know what the deal is with Unreal Engine, there's no charge until you get the first million dollars in revenue. Then the…”
   • [claimed-docs] “Access to the source code”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   No evidence in the pack addresses pricing/licensing history, terms stability, or retroactive repricing of shipped games; only a $15/month subscription mention and a link to the plans page appear, with no discussion of contract stability or track record.

   ⚑ flag

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

   weight 2 · round to PlayCanvas
   Unreal Engine~partialprobed5/10

   Unreal exposes substantial programmatic surfaces (Python editor scripting, C++ gameplay/engine API, Blueprints, command-line/UAT automation) that cover many editor and runtime tasks, but there is no documentation or evidence claiming full UI/API parity — some editor UI workflows (e.g. certain asset browser, viewport, or complex tooling operations) are not confirmed as scriptable, and the probe notes the engine is poorly machine-accessible overall (gated source repo, 403s on core domains, no official MCP, docs not served as markdown). missing for 10: an explicit statement/spec of full UI-API parity, evidence that every UI action has an API equivalent, and better machine-readable API discovery for AI agents.

   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [claimed-docs] “Automate time-consuming Asset management tasks in the Unreal Editor, like generating Levels of Detail (LODs) for Static Meshes.”
   • [claimed-docs] “there may be cases where you want precise control over when and how Assets are discovered, loaded, and audited. For these cases, the Asset M…”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [claimed-docs] “command-line arguments can be an invaluable tool for testing and optimizing your project”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialprobed6/10

   PlayCanvas exposes a scriptable Editor API (api.playcanvas.com/editor) for custom editing tools and integrations, plus an official MCP server that lets AI assistants modify project data and verify results via the viewport — going beyond a typical 'no API' game engine. However there is no discoverable REST/OpenAPI spec (404s across probed endpoints), and no documentation explicitly claims full UI-API parity; some Editor-only features (drag-and-drop scene building, real-time collaboration) are framed as UI-first conveniences without confirmed programmatic equivalents. missing for 10: an explicit parity statement/spec covering all UI actions, a formal REST/OpenAPI contract, and independent confirmation that collaborative/visual editor features are fully scriptable.

   • [claimed-docs] “The PlayCanvas Editor MCP Server connects AI assistants to an open PlayCanvas Editor session. It can modify project data and verify the resu…”
   • [claimed-docs] “This module is essential for developers looking to streamline their workflow, create custom editing tools, or integrate external data and sy…”
   • [probe] “official MCP server documented at https://developer.playcanvas.com/user-manual/editor/mcp-server/”
   • [probe] “PROBE runtime (recorded 2026-09-15): PlayCanvas ships the only OFFICIAL, open, npm-published editor MCP server of the eight engines — `npm v…”
   • [probe] “PROBE openapi: all candidate paths 404 (https://developer.playcanvas.com/openapi.json, https://developer.playcanvas.com/swagger.json, https:…”
   • [claimed-docs] “Want drag-and-drop scene building, real-time collaboration, or prefer visual tools over code”

   proof ↗⚑ flag

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

   weight 3 · round to PlayCanvas
   Unreal Engine—none0/10

   Evidence covers Blueprints, C++, plugins, source-code access, and asset management, but nothing addresses exporting a project's proprietary .uasset/.umap data into open interchange formats to allow a clean migration away from Unreal. The probe evidence even highlights poor machine-readable/open access to Epic's own ecosystem, reinforcing the lack of a data-portability story.

   • [claimed-docs] “provides a framework for managing Assets that can divide content into chunks that make sense in the context of your project”
   • [claimed-docs] “there may be cases where you want precise control over when and how Assets are discovered, loaded, and audited. For these cases, the Asset M…”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas~partialcommunity5/10

   The engine itself is open-source (MIT, npm-published) and PlayCanvas offers 'Download Apps for Self-hosting', plus community tooling (playcanvas-sync) to map project assets to a local filesystem, giving users a path to exit the platform with their code/engine. However, there is no first-party documentation of a comprehensive 'export all project data in open formats' feature, and community reports explicitly flag that code/collaboration data lives on PlayCanvas's servers with no easy local dev environment, undercutting a full data-portability story. Missing for 10: an official bulk data-export tool/format for scenes, assets and editor project state, and independent confirmation that exported data is fully open and complete.

   • [claimed-docs] “Download Apps for Self-hosting”
   • [community] “When I was working at a mobile game studio, we looked at this as an engine for a project. At the time we decided not to go with it since all…”
   • [community] “It's certainly not super obvious to setup but there are ways to use PlayCanvas with local setup... with some tinkering you can have hot modu…”
   • [community] “I recommend checking out the newly released playcanvas-sync tools which let you map your PlayCanvas project's assets to your local machine's…”
   • [community] “I think the engine is open source; but the real appeal is using the real time collab online tool, which is proprietary and had the drawbacks…”
   • [github] “npm install playcanvas”

   proof ↗⚑ flag

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

   weight 2 · round to PlayCanvas
   Unreal Engine!disputedcontradicted3/10

   Epic's own licensing page claims 'access to the source code' (unreal-docs-10, unreal-docs-18), but this is EULA/account-gated rather than an open-license grant — the probe confirms the GitHub source repo returns 404 without an authenticated, linked Epic account and all first-party domains block non-browser access, meaning the source is not freely readable under an open license as an AI-native user would need. Missing for 10: no OSI-approved or otherwise open license governing the source, no keyless/anonymous access to the code, no llms.txt or machine-readable confirmation of licensing terms.

   • [claimed-docs] “Access to the source code”
   • [claimed-docs] “Access to the source code All platforms & features Forums and documentation on the Epic Developer Community”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed9/10

   PlayCanvas Engine is confirmed open source (GitHub repo, MIT license, npm-published), and community commentary independently corroborates 'the engine is open source' (playcanvas-comm-11, comm-16). The full engine source is browsable on GitHub and versioned, satisfying an AI-native user's ability to read source under an open license; missing for 10: explicit license file text/OSI confirmation cited directly in evidence.

   • [github] “Gaussian Splatting - First-class support for loading and rendering 3D Gaussian Splats”
   • [github] “npm install playcanvas”
   • [community] “I think the engine is open source; but the real appeal is using the real time collab online tool, which is proprietary and had the drawbacks…”
   • [community] “It's an open source javascript game engine that is kinda like halfway between something like three.js and e.g. Unity—pretty lightweight, but…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

4. ai-native userSelf-host the core product

   weight 3 · round to Unreal Engine
   Unreal Engine~partialprobed6/10

   Unreal ships as a locally-installed engine with source-code access (unreal-docs-10/18), so the core product is inherently self-hostable rather than SaaS-only. However, the probe evidence shows source access is EULA/account-gated (must link a GitHub account to Epic), the GitHub repo is not publicly discoverable via API, and there is no simple open self-hosting path comparable to fully open-source engines. Missing for 10: unrestricted/public source access, no account-gating requirement, first-party documentation framing self-hosting as an explicit feature.

   • [claimed-docs] “Access to the source code”
   • [claimed-docs] “Access to the source code All platforms & features Forums and documentation on the Epic Developer Community”
   • [probe] “PROBE runtime negative (recorded 2026-09-15): Unreal is the least machine-readable presence of the eight — api.github.com/repos/EpicGames/Un…”

   proof ↗⚑ flag

   PlayCanvas!disputedcontradicted5/10

   The core Engine is open-source (MIT, npm-published) and documented for standalone/self-hosted use (playcanvas-docs-7, playcanvas-gh-2), but the actual PlayCanvas product most users mean—the Editor/collaboration platform—is proprietary SaaS with no self-hosting option; community reports confirm project code must live on PlayCanvas's servers with no easy local dev environment (playcanvas-comm-8, playcanvas-comm-11, playcanvas-comm-12). The 'Download Apps for Self-hosting' doc refers to exporting built apps, not self-hosting the Editor itself. Missing for 10: an official self-hostable Editor/backend, first-party self-host deployment docs for the full platform, and resolution of the community-confirmed server-lock-in for authoring.

   • [claimed-docs] “Build applications against the Engine directly. Learn how to use the Engine standalone.”
   • [github] “npm install playcanvas”
   • [claimed-docs] “Download Apps for Self-hosting”
   • [community] “When I was working at a mobile game studio, we looked at this as an engine for a project. At the time we decided not to go with it since all…”
   • [community] “I think the engine is open source; but the real appeal is using the real time collab online tool, which is proprietary and had the drawbacks…”
   • [community] “I'd love to be able to use their scene editor offline, uploading assets is a deal breaker.”

   proof ↗⚑ flag

Targets

1. studio leadOne project exports to desktop, mobile, and (directly or via partners) consoles

   weight 3 · round to Unreal Engine
   Unreal Engine~partialcommunity7/10

   Docs show a single project can be cooked/packaged/deployed to multiple target platforms via the cook/package pipeline (unreal-docs-16, unreal-docs-17) and tested across platforms via the toolbar (unreal-docs-9), and community mentions cross-platform export (e.g., HTML5) confirm this in practice (unreal-comm-14). However, the evidence pack contains no explicit documentation or independent confirmation of console export specifics (partner/licensee console SDK integration), so console support is inferred rather than directly evidenced. Missing for 10: explicit console-platform export/partner-program documentation, independent studio confirmation of a single project shipping to consoles.

   • [claimed-docs] “UnrealEditor.exe <GameName or uproject> -run=cook -targetplatform=<Plat1>+<Plat2>”
   • [claimed-docs] “This command can cook content for a platform, package it up into a platform's native format for distribution, deploys it to the device, and …”
   • [claimed-docs] “Use UE Toolbar to quickly test a part of your currently loaded Level for testing and debugging.”
   • [community] “For someone who's actually using UE4 as an artistic break for web dev, I thoroughly recommend it. It allows you to create great environments…”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   Evidence shows PlayCanvas engine and Editor covering web, desktop-via-browser, and self-hosted app export, but there is no mention anywhere in the pack of console export (PlayStation/Xbox/Switch) or partner porting programs. Community commentary consistently frames PlayCanvas as a web-first/WebGL engine with no console story. missing for 10: any documentation of console export or third-party console porting partnership, evidence of desktop/mobile packaging beyond generic 'download apps', console SDK support.

   • [claimed-docs] “Download Apps for Self-hosting”
   • [community] “It is a very nice engine, specially for those focused on Web only as target.”
   • [community] “It's an open source javascript game engine that is kinda like halfway between something like three.js and e.g. Unity—pretty lightweight, but…”

   proof ↗⚑ flag

Web

1. web developerMy game runs in the browser with a reasonable payload — WebGL today, and a credible WebGPU story

   weight 2 · round to PlayCanvas
   Unreal Engine—none0/10

   The only evidence touching web export is a decade-old community comment noting 'UE4.7 can export to HTML5 too' (unreal-comm-14) — there is no current documentation, no mention of WebGL, no mention of WebGPU, and no discussion of browser payload size or a credible modern web-export pipeline. Given Unreal actually removed HTML5/WebGL export in later versions, this stale anecdote does not establish a current capability.

   • [community] “For someone who's actually using UE4 as an artistic break for web dev, I thoroughly recommend it. It allows you to create great environments…”

   proof ↗⚑ flag

   PlayCanvas✓fullprobed8/10

   PlayCanvas is explicitly documented as a WebGL/WebGPU engine (probe-1) with a lightweight open-source runtime (npm playcanvas package, MIT license) and community corroboration that it's well-suited to lean web deployment (comm-1, comm-6, comm-13, comm-16) versus bloated alternatives. Missing for 10: no explicit payload-size benchmarks or bundle-size comparisons, and no independent hands-on confirmation of WebGPU parity/performance versus WebGL in production games.

   • [probe] “PROBE llms.txt: HTTP 200 at https://developer.playcanvas.com/llms.txt # PlayCanvas Developer Documentation > PlayCanvas is an open-source W…”
   • [github] “npm install playcanvas”
   • [community] “Unreal is a bloated way to develop WebGPU projects for the web. PlayCanvas is a sensible solution if your target is web.”
   • [community] “It is a very nice engine, specially for those focused on Web only as target.”
   • [community] “THREE.js is very low-level framework, while PlayCanvas additionally to low-level engine access has game framework, which will allow you to c…”
   • [community] “It's an open source javascript game engine that is kinda like halfway between something like three.js and e.g. Unity—pretty lightweight, but…”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”

   proof ↗⚑ flag

2. web developerThe engine installs from a package registry into my existing toolchain — a library I import, bundle, and tree-shake like any dependency

   weight 2 · round to PlayCanvas
   Unreal Engine—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

   PlayCanvas✓fullprobed8/10

   PlayCanvas engine is published to npm (`npm install playcanvas`, version 2.22.2, ~48k weekly downloads) as a standard ES module library, with docs on using the Engine standalone and scaffolding via create-playcanvas/Vite for standard bundler-based toolchains. Community confirms it integrates with other JS frameworks and modern web-dev workflows (HMR, bundlers), supporting tree-shakable ESM usage. Missing for 10: no explicit documentation or independent confirmation of tree-shaking behavior/bundle-size optimization guidance, and some community reports note setup isn't always straightforward.

   • [github] “npm install playcanvas”
   • [claimed-docs] “Build applications against the Engine directly. Learn how to use the Engine standalone.”
   • [claimed-docs] “Use `create-playcanvas` to scaffold a Vite and TypeScript project for the Engine, React, or Web Components from one of 12 runnable starters.”
   • [probe] “PROBE runtime (recorded 2026-09-15): developer.playcanvas.com serves llms.txt plus the largest llms-full.txt of the arena (2,053,637 bytes o…”
   • [community] “You can use it in conjunction with pretty much any other JS framework available.”
   • [community] “It's certainly not super obvious to setup but there are ways to use PlayCanvas with local setup... with some tinkering you can have hot modu…”

   proof ↗⚑ flag

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

   weight 2 · round drawn
   Unreal Engine—none0/10

   No evidence in the pack addresses telemetry, analytics, or usage-tracking opt-out settings for Unreal Engine or the Editor; all docs and community items concern scripting, licensing, and pricing topics unrelated to privacy controls.

   ⚑ flag

   PlayCanvas—none0/10

   No evidence pack item addresses telemetry, analytics, or opt-out settings for PlayCanvas engine, editor, or CLI tools; the axis applies (any dev tool/platform could ship telemetry controls) but there is simply no documentation or mention of such a setting.

   ⚑ flag

Iteration

1. game developerThe edit-run loop is fast — hot reload or near-instant preview after a script change

   weight 2 · round drawn
   Unreal Engine~partialclaimed6/10

   Docs confirm Blueprints allow instant in-editor scripting changes without recompilation, and C++ changes are 'reflected in the Unreal Editor after compiling' via Visual Studio/XCode, indicating a hot-reload/live-coding workflow exists. However, no evidence quantifies compile/reload speed, mentions the dedicated 'Live Coding' feature explicitly, or provides independent hands-on confirmation that the loop is fast rather than just functional. Missing for 10: benchmark/timing evidence on hot reload speed, explicit mention of Live Coding feature, independent developer testimony on edit-run loop responsiveness.

   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”

   proof ↗⚑ flag

   PlayCanvas~partialcommunity6/10

   Docs claim live editing lets you tweak materials, positions, and even modify scripts while the application is running, and community reports confirm hot-module-replacement style workflows are achievable with the local sync tooling — but multiple community sources note this setup is 'certainly not super obvious' and historically required 'tinkering,' with one report citing no easy local dev environment as a dealbreaker at the time. missing for 10: first-party benchmark or docs describing built-in near-instant reload latency, and confirmation the friction reported in community threads has been resolved.

   • [claimed-docs] “With live editing, you can tweak materials, adjust positions, and even modify scripts while your application runs.”
   • [community] “It's certainly not super obvious to setup but there are ways to use PlayCanvas with local setup... with some tinkering you can have hot modu…”
   • [community] “When I was working at a mobile game studio, we looked at this as an engine for a project. At the time we decided not to go with it since all…”
   • [community] “I recommend checking out the newly released playcanvas-sync tools which let you map your PlayCanvas project's assets to your local machine's…”

   proof ↗⚑ flag

Language

1. game developerThe primary scripting language is productive and fully exposes the engine API, with a debugger behind it

   weight 3 · round to PlayCanvas
   Unreal Engine~partialcommunity5/10

   Docs show C++ as the primary language with full engine API exposure (new Gameplay classes, delegates binding to arbitrary objects) and it compiles/reflects via Visual Studio/XCode, implying standard IDE debugger support, while Blueprints (the more 'scripting'-like option) explicitly trades off some API depth for accessibility (unreal-docs-1,2,3,11). Community evidence corroborates a productivity/API-completeness tension: Blueprints alone are 'not sufficient' for real scripting tasks and require C++ fallback, and multiple developers cite C++ as their 'main problem' with UE (unreal-comm-4, unreal-comm-7, unreal-comm-15). Missing for 10: no explicit documentation of an integrated scripting debugger workflow (only external Visual Studio/Xcode debugging is implied), and no first-party evidence that any single primary language achieves both full productivity and full API coverage without trade-offs.

   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [claimed-docs] “You can dynamically bind a delegate to a member function of an arbitrary object, calling the function on the object at a future time”
   • [claimed-docs] “Creating new Gameplay classes in C++, and all changes will be reflected in the Unreal Editor after compiling with either Visual Studio or XC…”
   • [community] “With no scripting support, adding content like vehicles is pretty much impossible. Blueprints are not sufficient for such a task. A few peop…”
   • [community] “My main problem with UE is C++, I don't want to even try to download files or read json in a blueprint.”
   • [community] “blueprints are great for prototyping, but once you have too much complexity in a BP it can easily cascade inefficiencies to the point it bec…”

   proof ↗⚑ flag

   PlayCanvas~partialcommunity6/10

   PlayCanvas scripting uses standard JavaScript/TypeScript (ESM `.mjs` scripts with class syntax) and the full engine can be scripted via `npm install playcanvas`, giving broad API access and a productive, familiar language; community comments (e.g. comm-13) also praise its game-framework layer atop low-level access. However, the evidence pack never explicitly documents a debugger (e.g. browser devtools breakpoints, source maps, or an IDE debug workflow) tied to scripting, and some community feedback flags dev-experience friction (playcanvas-comm-2, playcanvas-comm-3, playcanvas-comm-9) around local/offline workflows that could affect debugging setup. Missing for 10: explicit documentation of a debugger/breakpoint workflow for scripts, and confirmation that the full engine API (not just scripting hooks) is exposed without gaps.

   • [claimed-docs] “ESM Scripts (`.mjs` files) — Modern ES Module-based scripts using class syntax. Recommended for new projects.”
   • [claimed-docs] “ESM Scripts (`.mjs` files) — Modern ES Module-based scripts using class syntax. **Recommended for new projects.**”
   • [claimed-docs] “Build applications against the Engine directly. Learn how to use the Engine standalone.”
   • [github] “npm install playcanvas”
   • [community] “THREE.js is very low-level framework, while PlayCanvas additionally to low-level engine access has game framework, which will allow you to c…”
   • [community] “It's certainly not super obvious to setup but there are ways to use PlayCanvas with local setup... with some tinkering you can have hot modu…”

   proof ↗⚑ flag

Visual

1. technical artistBuild gameplay logic with visual scripting without writing code

   weight 2 · round to Unreal Engine
   Unreal Engine!disputedcontradicted6/10

   Epic's docs confirm Blueprints let designers use most programmer-level tools without code (unreal-docs-1), and this is Unreal's flagship visual-scripting system for gameplay logic. However, hands-on community reports concretely contradict full no-code sufficiency: one dev states Blueprints were 'not sufficient' for adding content like vehicles and required C++ (unreal-comm-4), and another notes complex Blueprint graphs 'cascade inefficiencies' making C++ necessary at scale (unreal-comm-15), with a third citing C++ as the main pain point when trying to avoid it (unreal-comm-7). Missing for 10: independent case studies of large shipped games built purely in Blueprints without C++ fallback, and clearer scoping of what gameplay complexity Blueprints alone can handle.

   • [claimed-docs] “provides the ability for designers to use virtually the full range of concepts and tools generally only available to programmers”
   • [community] “With no scripting support, adding content like vehicles is pretty much impossible. Blueprints are not sufficient for such a task. A few peop…”
   • [community] “blueprints are great for prototyping, but once you have too much complexity in a BP it can easily cascade inefficiencies to the point it bec…”
   • [community] “My main problem with UE is C++, I don't want to even try to download files or read json in a blueprint.”

   proof ↗⚑ flag

   PlayCanvas—none0/10

   Evidence shows PlayCanvas scripting is code-based (ESM/.mjs JavaScript classes) and the Editor supports drag-and-drop scene building and collaboration, but there is no mention of a node-based or visual scripting system for gameplay logic comparable to Blueprints/Bolt that would let a technical artist avoid writing code.

   • [claimed-docs] “ESM Scripts (`.mjs` files) — Modern ES Module-based scripts using class syntax. Recommended for new projects.”
   • [claimed-docs] “Want drag-and-drop scene building, real-time collaboration, or prefer visual tools over code”
   • [claimed-docs] “ESM Scripts (`.mjs` files) — Modern ES Module-based scripts using class syntax. **Recommended for new projects.**”

   proof ↗⚑ flag