feat(m3): Epic MCP scaffold + live probe -> retirement matrix (52 toolsets / 830 tools) - #14
Conversation
- docs/EPIC_MCP_MATRIX.md: the retirement contract-of-record. Live-probed Epic official Unreal MCP (UE 5.8.0, AllToolsets): 52 toolsets / 830 tools, full schemas captured. Verdicts for all 58 of our tools: 36 RETIRE (behind legacy flag), 17 KEEP, 5 KEEP-PARTIAL — every claim cites exact probe tool names; absences verified by whole-probe search. Key facts: Tool Search Mode is default (tools/list = 3 meta-tools), execute_tool_script is a sandboxed batcher (no unreal import — our ue_execute_python stays unique), no console exec / undo / lighting / cloner / continuous-perception anywhere in the surface. - docs/epic_mcp/: raw probe captures (AllToolsets + blank-default) and the Programmatic exec-sandbox environment record. - scripts/probe_epic_mcp.py: repeatable prober (rerun per engine version; the diff is the drift report). - .mcp.json: two-server config — our stdio bridge + Epic http :8000. - .uproject: ModelContextProtocol + AllToolsets as Optional:true (no-ops on 5.7, self-enables when the project opens in 5.8). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- ue_query_scene row: find_actors returns refs only; location-bearing results cost O(N) transform follow-ups (reopen condition added) - snap-to-ground negative claim scoped: spawn-time snap_to_ground args exist; no standalone tool for existing actors, no normal output - legend: add PrimitiveTools prefix Verify wave confirmed: all 26 spot-checked Epic citations exist verbatim, all 4 negative claims independently reproduced, headline arithmetic exact. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Warning Review limit reached
Next review available in: 59 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (7)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
1 participant
What
The M3 milestone: Epic's official Unreal MCP (UE 5.8) live-probed and turned into the retirement contract-of-record.
docs/EPIC_MCP_MATRIX.md— dispositions for all 58 of our tools against Epic's real surface: 36 RETIRE (behind the M4 legacy flag), 17 KEEP, 5 KEEP-PARTIAL. Every claim cites exact probe tool names; every absence claim (no console exec, no undo, no lighting orchestration, no cloner, no continuous capture, no arbitrary Python) verified by whole-capture search.docs/epic_mcp/— raw captures: full AllToolsets surface (52 toolsets / 830 tools with schemas), blank-project default (AgentSkillToolset only), and theexecute_tool_scriptsandbox record (module allowlist has nounreal— ourue_execute_pythonstays unique).scripts/probe_epic_mcp.py— rerun per engine version; the diff is the API-drift report..mcp.json— two-server config (our stdio bridge +unreal-epichttp :8000)..uproject—ModelContextProtocol+AllToolsetsasOptional: true(no-op on 5.7, self-enables on 5.8).Key findings
Epic's MCP plugin ships zero tools — the surface lives in
Toolsets/*plugins. Default mode is Tool Search (3 meta-tools; concrete tools viacall_tooldispatch;bEnableToolSearch=falseflips them first-class). Tool calls run serially on the game thread. EULA §6(e) governs LLM data flow.Verification
Adversarial audit: all 26 spot-checked citations exist verbatim in the capture; all 4 negative claims independently reproduced; headline arithmetic recounted row-by-row.
🤖 Generated with Claude Code