skill-candidates/repo-b-mcp-comfy-bridge/SKILL.md
Canonical MCP adapter and Comfy bridge operations for <PRIVATE_REPO_B>. Use when enabling MCP, validating shim.comfy resources/tools, operating workflow/pipeline submissions, or running fail-closed Comfy diagnostics with optional AMUSE and CapCut contract checks.
npx skillsauth add grtninja/skill-arbiter repo-b-mcp-comfy-bridgeInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Use this skill as the canonical MCP + Comfy lane in <PRIVATE_REPO_B>.
A legacy local Comfy wrapper is retained for backward-compatible triggers and should route here for new work.
/api/mcp/status./api/mcp/config changes only when configuration drift is confirmed.shim.comfy.status, shim.comfy.queue, shim.comfy.history).shim.comfy.prompt.submit, shim.comfy.workflow.submit, shim.comfy.pipeline.run).Use this skill for MCP adapter state/config plus shim.comfy.* resource health.
Do not use this skill for:
$env:SHIM_ENABLE_MCP = "1"
$env:SHIM_MCP_HOST = "127.0.0.1"
$env:SHIM_MCP_PORT = "9550"
$env:SHIM_MCP_ALLOW_LAN = "0"
$env:MX3_COMFYUI_ENABLED = "1"
$env:MX3_COMFYUI_BASE_URL = "http://127.0.0.1:8188"
$env:MX3_COMFYUI_TIMEOUT_S = "10"
$env:MX3_COMFYUI_DEFAULT_WORKFLOW_PROFILE = "small_video"
$env:MX3_AMUSE_ENABLED = "1"
$env:MX3_AMUSE_BASE_URL = "http://127.0.0.1:3001"
$env:MX3_AMUSE_TIMEOUT_S = "15"
$env:REPO_B_LOCAL_COMFY_ORCH_FAIL_CLOSED = "1"
curl http://127.0.0.1:9000/api/mcp/status
curl http://127.0.0.1:9000/api/comfy/workflows/templates
curl http://127.0.0.1:9000/api/comfy/pipelines/profiles
curl http://127.0.0.1:9000/api/amuse/status
curl http://127.0.0.1:9000/api/amuse/capabilities
Apply config:
curl -X POST http://127.0.0.1:9000/api/mcp/config \
-H "content-type: application/json" \
-d '{"enabled":true,"persist":true,"apply_now":true}'
Validate these resource contracts before accepting diagnostics:
shim.comfy.status:
enabled=truereachable=truelast_errorchecked_atshim.comfy.queue:
running_prompt_ids and pending_prompt_ids listsrunning_count and pending_count must match list lengthsshim.comfy.history:
entries listcount must match entries lengthshim.comfy.workflow.submit accepts one of:
workflow objectworkflow_idworkflow_pathworkflow_profileshim.comfy.pipeline.run supports profile defaults and optional AMUSE stage:
profile values include small_video_capcut and quality_video_capcutwait_for_state must normalize to NONE|QUEUED|RUNNING|DONEcapcut_preset=true, response includes capcut_export.editor=capcutamuse_enhance=true, response includes additive amuse payloadPipeline run example:
curl -X POST http://127.0.0.1:9000/api/comfy/pipelines/run \
-H "content-type: application/json" \
-d '{"profile":"small_video_capcut","wait_for_state":"RUNNING","capcut_preset":true}'
When deterministic artifact output is needed, use the legacy local orchestrator drop-ins
(tools/local_comfy_orchestrator.py, tools/local_comfy_validate.py):
python tools/local_comfy_orchestrator.py \
--task comfy-health-check \
--json-out .codex/local_comfy/comfy-health-check.json \
--repo-root .
MX3_COMFYUI_ENABLEDMX3_COMFYUI_BASE_URLMX3_COMFYUI_API_KEYMX3_COMFYUI_TIMEOUT_SMX3_COMFYUI_DEFAULT_WORKFLOW_PROFILEMX3_AMUSE_ENABLEDMX3_AMUSE_BASE_URLMX3_AMUSE_TIMEOUT_Sreferences/mcp-comfy-checklist.mdIf this lane is unresolved, blocked, or ambiguous:
$skill-hub for chain recalculation.tools
Run a defender-first security sweep on code, configs, prompts, model/tooling surfaces, or third-party contribution lanes. Use when a request involves safe bug, leak, zero-day-class, exploit, or hack hunting for protection, when contributing to outside repositories and you want a focused security pass, or when touching auth, secrets, permissions, network exposure, prompt/tool boundaries, data flow, or update/build surfaces. This skill is defensive only and must never be used for weaponization or unauthorized access.
development
Validate and repair VRM Sandbox startup acceptance with shim-first local model authority, frontend/backend bring-up, and avatar-runtime launch proof. Use when launch behavior, chat handoff, voice fallback, or runtime bridge acceptance must be verified end to end.
documentation
Align documented voice-command catalogs, endpoint action allowances, and live runtime handlers so operator-visible voice surfaces match what the stack can actually execute. Use when voice command docs, parser matrices, endpoint permissions, or runtime action routing drift apart.
development
Track SkillHub trend and topic drift, maintain a bounded rewrite watchlist, and surface emerging gaps worth turning into repo-owned skills. Use when the marketplace query set shows new families or when the current shortlist has gone stale.