본문으로 건너뛰기
cccskills
무료GitHub에 공개

agenttrail

agenttrail: live map of a multi-agent build in the browser: which plan component is being worked on, by which agent or harness, what is done and what is stuck. Use when a multi-agent pipeline starts (/startcycle, /startcycle-graph, /teamwork-preview) or after a plan-canvas approve, when the user asks to see what the agents are doing ("live map", "build board", "who is running now"), or by yourself whenever a multi-agent run is under way — no user prompt needed.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

설치 방법 보기

포함된 파일(26)

  • SKILL.md5.0 KB
  • bin/agenttrail.mjs70.2 KB
  • bin/ensure.mjs8.2 KB
  • bin/repoid.mjs3.8 KB
  • bin/trail-hub.mjs15.8 KB
  • lib/click.mjs15.4 KB
  • lib/code-stamp.mjs2.0 KB
  • lib/events.mjs8.0 KB
  • lib/gatepanel.mjs6.7 KB
  • lib/httpguard.mjs2.3 KB
  • lib/hub-model.mjs11.3 KB
  • lib/hub-state.mjs2.3 KB
  • lib/identity.mjs3.4 KB
  • lib/plan.mjs4.4 KB
  • lib/sessions.mjs7.5 KB
  • lib/shared-static.mjs1.6 KB
  • lib/stale.mjs3.2 KB
  • LICENSE1.0 KB
  • public/hub/app.css5.8 KB
  • public/hub/app.js22.5 KB
  • public/hub/enroll.html923 B
  • public/hub/enroll.js2.0 KB
  • public/hub/index.html2.3 KB
  • public/index.html34.9 KB
  • public/shared/map.css49.6 KB
  • public/shared/map.js45.6 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

<!-- Source: sodiumsun/agenttrail bin/agenttrail.mjs + public/index.html — MIT, see THIRD_PARTY_NOTICES.md -->

agenttrail — live pipeline map

agenttrail serves a browser board that mirrors a multi-agent build in real time: which plan component is being worked on, by which agent or harness, what is done and what is stuck. It observes only — it never runs agents and never marks tasks done.

Start the map

aos-trail . --plan production_artifacts/00_execution_plan.md --no-open

It prints the URL (default http://localhost:5330, next free port if taken). Open that URL for the user. If running inside AO (env var AO_BROWSER_CAPABILITY is set), run ao preview <url> so it shows in AO's Browser tab. Without a plan file, start it with just aos-trail . — it then shows file activity only.

Ensure (auto-start)

aos-trail --ensure [--cwd <dir>] [--plan <file>] [--session <id>] [--json]

Starts the map detached if none runs for this repo (matched via /whoami repoPath on 127.0.0.1:5330-5344). repoPath is the repo's main checkout, so every git worktree of a repo finds the same map; the map shows each worktree as its own lane (branch and folder) and files hook events by their cwd. /whoami also lists worktrees, opens it at most once per session (state in $TMPDIR/aos-trail-ensure/), and always exits 0. Plan: --plan, else production_artifacts/00_execution_plan.md, else the single production_artifacts/*/00_execution_plan.md; it needs a {#id} marker. No open under CI, SSH, or headless Linux; with AO_BROWSER_CAPABILITY it runs ao preview <url>. --json prints {url,started,opened,reason,plan,hint}. Test overrides: AOS_TRAIL_OPENER (opener command), AOS_TRAIL_PORTS=lo-hi (probe range).

Lanes: main first, then at most 11 worktrees; the slots go to the most recently active ones (a lane with events in the last 10 min keeps its slot), and events from any further worktree of the repo are filed in one other worktrees lane, never in main. /whoami carries protocol: 2. --ensure replaces a daemon on this repo (or one of its worktrees) that lacks it: the new map starts first, then the old one is asked to quit via POST /shutdown (JSON body naming its own repoPath); a daemon that cannot (older ones have no such API) keeps running and hint says which port to stop. Two concurrent --ensure runs share a lock file in $TMPDIR/aos-trail-ensure/. /setup on a map with several lanes needs {"lane": "<key>"} and writes into that checkout.

Plan convention

The plan file uses components and tasks:

  • ## Plain-language name {#id} = a component (5-9 per plan), stable ids
  • under it optional lines: needs: [id, id], links: [id], files: [src/**], url: production_artifacts/00_architecture.html
  • tasks: - [ ] Outcome {#task-id}; mark [~] BEFORE starting, [x] when done, [!] when stuck, and add an indented by: <agent> line (claude, codex, agy, opencode)
  • save the file immediately after each status change; never batch updates to the end
  • Mermaid blocks and prose may stay in the file; the map ignores them.

Live agent events

The AOS installer registers .claude/hooks/trail-relay.mjs --agent <harness> for Claude Code (PreToolUse, PostToolUse, SessionStart, Stop, SubagentStop), Antigravity and Codex (PreToolUse, Stop). The OpenCode plugin and mcsc (for the CLI workers it delegates to) post events directly. Events are fire-and-forget (300 ms cap) and never block a tool call; with no map running they are dropped.

Automatic node link. A dispatcher sets AOS_TRAIL_NODE=<node-id> and AOS_TRAIL_PLAN=<plan-file> when it starts a worker (aos-acp --trail-node <id> --trail-plan <file> does this). The relay stamps both onto every event, so the node is live on the map while the worker reports and is shown as touched, not done, after its Stop or SessionEnd. Interactive sessions: aos-trail emit task.update --task <id> --status doing --plan <file>.

A component's url: may be a path under production_artifacts/ — the map serves that folder, so e.g. url: production_artifacts/00_architecture.html opens the archify diagram from the card.

Stop

The daemon runs until its terminal/process ends; aos-trail up relaunches saved boards.

Rules

  • Never create a root PLAN.md in a user repo during an AOS pipeline — always point at the pipeline's plan with --plan.
  • Never mark [x] without evidence in the code.

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

Use when manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Use when a hybrid memory system that provides persistent, searchable knowledge management for AI agents (Architecture, Patterns, Decisions).

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Reference for how BDB structures autonomous software engineering work — the seven-node dispatcher graph (Architect, TechLead, UI/UX, Engineering, Media/EventTech, Reviewer, Shipping) that /startcycle-graph actually runs. Use when you need the high-level lifecycle framing without inventing your own process.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Tools are how AI agents interact with the world. A well-designed tool is the difference between an agent that works and one that hallucinates, fails silently, or costs 10x more tokens than necessary. This skill covers tool design from schema to error handling.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Harness patterns for coding agents — memory, permissions, context engineering, delegation, skills, hooks, bootstrap.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

AI agent development workflow for building autonomous agents, multi-agent systems, and agent orchestration with CrewAI, LangGraph, and custom agents.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

hybridlabor-api의 스킬 모두 보기

이 스킬의 문제 신고