Skip to content

docs(skills): a0-archon-bridge — align with live v2.11 surface + fix :8091 (supersedes #2865) - #2942

Merged
POWERFULMOVES merged 3 commits into
mainfrom
feat/a0-archon-bridge-skill
Sep 8, 2026
Merged

POWERFULMOVES merged 3 commits into
mainfrom
feat/a0-archon-bridge-skill

Conversation

@POWERFULMOVES

Copy link
Copy Markdown
Owner

Supersedes #2865 (its head branch was deleted from the remote and the PR would not rebind; content is rebuilt as a single commit on current main — see the close comment there).

What

Aligns .claude/skills/a0-archon-bridge/SKILL.md with the live Agent Zero v2.11-hardened surface (5c280a9) and corrects the Archon port map.

  • Handler inventory: 33 features on v2.11 (message_queue, browser_runtime_config, launcher_gateway_file_write, remote tool handlers, compact_chat, …) + the capabilities payload's auth/streaming/websocket_namespace/attachments contract
  • MCP arg filter (post-chore(submodules): promote PMOVES-Agent-Zero to 5c280a9 (MCP arg filter) #2852): filter_declared_args drops schema-undeclared keys at the client boundary — orange log, not rejection
  • Wrapper env (post-fix(agent-zero): configurable message timeout, POST health probe, honest healthz + OTEL parity #2813): AGENT_ZERO_HEALTH_METHOD=POST (capabilities is POST-only; GET 405s), AGENT_ZERO_MESSAGE_TIMEOUT=600 (was hardcoded 60s)
  • healthz is a child-process check, not a reachability check (B850 measurement 2026-09-02): 200 + runtime.note "health endpoint not found (404)" means wrapper up, connector path not — read runtime.status AND runtime.note, not the HTTP code
  • :8091 corrected: one Archon service on container :3090, host-published by default on :8091 (ARCHON_API_PORT, compose agents overlay) with :3737 a UI-compat alias — identical /api/health measured on all three (SPARK, 2026-09-04). mcp-gateway's in-network 8091 is host-published 8189 by default. Replaces feat(skills): a0-archon-bridge — documented integration points + live NATS/CHIT state #2783's "probing :8091 reaches nothing" warning, which contradicted both the compose default and the live fleet

Testing

  • Live probes (SPARK, 2026-09-04): :3090/:3737/:8091 /api/health → identical Archon payload; docker ps shows all three published onto pmoves-archon-1's 3090
  • Compose on main: docker-compose.agents.yml:374 ${ARCHON_API_PORT:-8091}:3090, :373 UI-compat comment + NEXT_PUBLIC_ARCHON_UI_URL default; mcp-gateway overlay documents the collision and its 8189 host default
  • healthz behavior citations: pmoves/services/agent-zero/main.py:843-858, :331-338, :354-360

…:8091

Rebuild of the #2865 lane onto current main (the remote branch was
deleted; local checkout was also two commits behind the PR head, which is
folded into this single commit):

- Handler inventory: 33 features live on v2.11-hardened 5c280a9
  (message_queue, browser_runtime_config, launcher_gateway_file_write,
  remote tool handlers, compact_chat...), plus the capabilities payload's
  auth/streaming/websocket-namespace/attachments contract.
- MCP arg filter: filter_declared_args drops schema-undeclared keys at
  the client boundary (orange log, not rejection) — post-#2852.
- Wrapper env: post-#2813 additions AGENT_ZERO_HEALTH_METHOD=POST
  (capabilities is POST-only; GET 405s) and
  AGENT_ZERO_MESSAGE_TIMEOUT=600 (was hardcoded 60s).
- healthz is a child-process check, not a reachability check (B850
  measurement 2026-09-02): 200 with runtime.note "health endpoint not
  found (404)" means the wrapper is up and the connector path is not —
  read runtime.status AND runtime.note, not the HTTP code.
- :8091 corrected: one Archon service on container :3090, host-published
  by default on :8091 (ARCHON_API_PORT, compose agents overlay) with
  :3737 alias — measured identical /api/health on all three (SPARK
  2026-09-04). mcp-gateway's in-network 8091 is host-published 8189 by
  default. Replaces #2783's "probing :8091 reaches nothing" warning,
  which contradicted both the compose default and the live fleet.

Rebasing onto current main also refreshes the PMOVES-Agent-Zero gitlink
past the pre-v2.11 pin that failed agent-zero-pin-check in CI.

Co-Authored-By: B850 Claude (pmoves-b850) <b850@pmoves.ai>
@coderabbitai

coderabbitai Bot commented Sep 4, 2026 •

Copy link
Copy Markdown
Contributor

Important

Review skipped

Auto reviews are limited based on label configuration.

🏷️ Required labels (at least one) (1)
  • coderabbit

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Advanced

Run ID: ada562c2-fbc4-40d0-9d50-cf232cc40c62

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

  • 🔍 Trigger review

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.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Sep 4, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-09-04T22:03:29.381497Z 27582cd PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@github-actions github-actions Bot added the docs Documentation label Sep 4, 2026

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 27582cdd1e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread .claude/skills/a0-archon-bridge/SKILL.md
@POWERFULMOVES

Copy link
Copy Markdown
Owner Author

Review request → 4090-CLAUDE (operator-authorized, waiting to sign)

The one gap on this PR is the peer approval: claude-review check is green, Codex reviewed clean, all CI green — but no approving review yet. Per the routing this is 4090's slot.

What's here for the reviewer, in order:

Dispatch note: attempted pmoves.agent.task.v1 publish from SPARK — NATS unreachable from this session's MCP (no creds wired; #2937 is that lane). This comment is the durable request surface.

POWERFULMOVES pushed a commit that referenced this pull request Sep 7, 2026
…n 404

P2 review finding on #2942: the validation command printed only the
top-level status, which reads "ok" in exactly the failure mode the skill
warns about (connector 404 swallowed into a 200). The one-liner now
checks runtime.status and rejects a 404 note, exiting nonzero on an
unreachable connector.

💘 Generated with Crush
…n 404

P2 review finding on #2942: the validation command printed only the
top-level status, which reads "ok" in exactly the failure mode the skill
warns about (connector 404 swallowed into a 200). The one-liner now
checks runtime.status and rejects a 404 note, exiting nonzero on an
unreachable connector.

💘 Generated with Crush
@POWERFULMOVES
POWERFULMOVES force-pushed the feat/a0-archon-bridge-skill branch from 652d313 to b8a78ee Compare September 7, 2026 12:48
@POWERFULMOVES

Copy link
Copy Markdown
Owner Author

◇ Addressed in b8a78ee80 — the healthz one-liner now inspects runtime.status + runtime.note, exits nonzero on a 404 note, and prints an explicit HEALTHY/UNREACHABLE-CONNECTOR verdict.

POWERFULMOVES pushed a commit that referenced this pull request Sep 8, 2026
…ills PRs, on-node passes

Adapts pmoves-pair-review and PAIR_REVIEW_RECIPROCITY.md from
parallel-CLAUDE-only to the fleet as it actually is (operator-directed,
during the #2943 review session):

- Any harness counts as the peer surface: Claude Code, Crush, Codex CLI,
  Hermes, KiloCode claws, DeepSeek — cross-harness reviews are the point,
  they surface harness-assumption drift (env vars one launcher sources
  and another doesn't; the TS_Z890 roster drop was this class)
- Fourth reviewer surface: the operator (DARKXSIDE/POWERFULMOVES) as
  Control — both #2942's :8091 correction and #2938's topology challenge
  came from operator pushback on confident agent drafts; gate, don't
  bottleneck
- Harness notes for Crush: same gh/AGNOTE/sign-trail mechanics,
  COMMENTED-not-APPROVED (same-account constraint), sign with the
  agent_signatures.yaml identity, skills-load-on-demand as a trigger-
  quality check
- Skills-PR review checklist mapped to the 4-class taxonomy: frontmatter
  name==dir (Crush validator), description-is-a-trigger, anchors ratchet,
  live-surface probes for port/endpoint claims, Tier 1/Tier 2 CI
- On-node/private review passes: Ollama local models (private by
  construction — motivated by the 2026-09-04 tailnet-address leak),
  hf-agent model nomination, Archon 0.6.0 conversation endpoints for
  tool-using passes, and the already-provisioned coding plans via their
  CLIs — with the leads-not-verdicts guardrail

No behavior change — workflow docs + skill only.
POWERFULMOVES pushed a commit that referenced this pull request Sep 8, 2026
…ills PRs, on-node passes

Adapts pmoves-pair-review and PAIR_REVIEW_RECIPROCITY.md from
parallel-CLAUDE-only to the fleet as it actually is (operator-directed,
during the #2943 review session):

- Any harness counts as the peer surface: Claude Code, Crush, Codex CLI,
  Hermes, KiloCode claws, DeepSeek — cross-harness reviews are the point,
  they surface harness-assumption drift (env vars one launcher sources
  and another doesn't; the TS_Z890 roster drop was this class)
- Fourth reviewer surface: the operator (DARKXSIDE/POWERFULMOVES) as
  Control — both #2942's :8091 correction and #2938's topology challenge
  came from operator pushback on confident agent drafts; gate, don't
  bottleneck
- Harness notes for Crush: same gh/AGNOTE/sign-trail mechanics,
  COMMENTED-not-APPROVED (same-account constraint), sign with the
  agent_signatures.yaml identity, skills-load-on-demand as a trigger-
  quality check
- Skills-PR review checklist mapped to the 4-class taxonomy: frontmatter
  name==dir (Crush validator), description-is-a-trigger, anchors ratchet,
  live-surface probes for port/endpoint claims, Tier 1/Tier 2 CI
- On-node/private review passes: Ollama local models (private by
  construction — motivated by the 2026-09-04 tailnet-address leak),
  hf-agent model nomination, Archon 0.6.0 conversation endpoints for
  tool-using passes, and the already-provisioned coding plans via their
  CLIs — with the leads-not-verdicts guardrail

No behavior change — workflow docs + skill only.
POWERFULMOVES added a commit that referenced this pull request Sep 8, 2026
…ills PRs, on-node passes (#2946)

* docs(skills): pair-review goes harness-agnostic — Crush, operator, skills PRs, on-node passes

Adapts pmoves-pair-review and PAIR_REVIEW_RECIPROCITY.md from
parallel-CLAUDE-only to the fleet as it actually is (operator-directed,
during the #2943 review session):

- Any harness counts as the peer surface: Claude Code, Crush, Codex CLI,
  Hermes, KiloCode claws, DeepSeek — cross-harness reviews are the point,
  they surface harness-assumption drift (env vars one launcher sources
  and another doesn't; the TS_Z890 roster drop was this class)
- Fourth reviewer surface: the operator (DARKXSIDE/POWERFULMOVES) as
  Control — both #2942's :8091 correction and #2938's topology challenge
  came from operator pushback on confident agent drafts; gate, don't
  bottleneck
- Harness notes for Crush: same gh/AGNOTE/sign-trail mechanics,
  COMMENTED-not-APPROVED (same-account constraint), sign with the
  agent_signatures.yaml identity, skills-load-on-demand as a trigger-
  quality check
- Skills-PR review checklist mapped to the 4-class taxonomy: frontmatter
  name==dir (Crush validator), description-is-a-trigger, anchors ratchet,
  live-surface probes for port/endpoint claims, Tier 1/Tier 2 CI
- On-node/private review passes: Ollama local models (private by
  construction — motivated by the 2026-09-04 tailnet-address leak),
  hf-agent model nomination, Archon 0.6.0 conversation endpoints for
  tool-using passes, and the already-provisioned coding plans via their
  CLIs — with the leads-not-verdicts guardrail

No behavior change — workflow docs + skill only.

* fix(pair-review): five review-finding corrections to skill + reciprocity doc

- SKILL.md: drop the stale CLAUDE-only "do NOT invoke" block that
  contradicted the harness-agnostic one; operator signs as darkxside/
  powerfulmoves (dsh is not a registered signature identity)
- RECIPROCITY.md: hf-agent is an autonomous HF patrol with no per-diff
  endpoint — reframe as a discovery feed, selection stays with the
  session; Archon is not on every node — probe /api/health (JSON body)
  before offering it; coding-plan credentials land per-harness tiers
  (CLAUDE_CODE_OAUTH_TOKEN registers in the agent tier), resolve from
  the secrets manifest instead of assuming env.tier-llm

💘 Generated with Crush

---------

Co-authored-by: Agent Zero <agent.zero@pmoves.ai>
@POWERFULMOVES
POWERFULMOVES merged commit 70b7fd2 into main Sep 8, 2026
26 checks passed
@POWERFULMOVES
POWERFULMOVES deleted the feat/a0-archon-bridge-skill branch September 8, 2026 10:03
POWERFULMOVES added a commit that referenced this pull request Sep 8, 2026
…dy ACK (#2998)

* docs(register): five-PR merge wave closeout — RELEASE rows + three-body ACK

RELEASE rows for archon-native (#2943), pair-review (#2946),
a0-archon-bridge (#2942), gh-app mint (#2959), H3 V3 archive (#2980).
Signoff checklist carries the wave's delivery/control/memory ACKs;
control is the operator's in-session merge directive. Learnings artifacts
stayed runtime-generated per protocol.

💘 Generated with Crush

* fix(register): declare CRUSH-SPARK (KIMI) and the kimi-k2 model token

The 2026-09-08 RELEASE rows signed as CRUSH-SPARK (KIMI) -- the model+node
decoration shape the vocabulary already documents for CRUSH-GLM52 (Z890).
Alias folded to the crush identity; KIMI declared as the kimi-k2 model.

💘 Generated with Crush

---------

Co-authored-by: Agent Zero <agent.zero@pmoves.ai>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

docs Documentation

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant