AgentSkillsCN

samocode-run

在项目中运行并监控samocode自主会话。当用户说“运行samocode”或希望继续samocode会话时使用。

SKILL.md
--- frontmatter
name: samocode-run
description: Run and monitor samocode autonomous sessions on projects. Use when user says "run samocode" or wants to continue a samocode session.

Samocode Run

Runs the samocode autonomous orchestrator on a project session and monitors its progress.

CRITICAL: DO NOT MANUALLY ORCHESTRATE

When user asks to "run samocode" or "continue samocode", you MUST use this skill.

DO NOT:

  • Launch Task subagents yourself for investigation/planning/implementation phases
  • Manually read _overview.md and decide what phase to run
  • Update _signal.json yourself
  • Pretend to be the orchestrator

The Python worker (main.py) handles ALL of this. Your job is to START the worker and MONITOR its output.

Trigger Phrases

Use this skill when user says:

  • "run samocode"
  • "start samocode"
  • "continue samocode"
  • "let samocode work on it"

What is Samocode?

Samocode is an autonomous session orchestrator that runs Claude CLI in a loop to complete complex tasks. It:

  • Reads session state from _overview.md
  • Runs phase-specific agents automatically based on current phase
  • Sends Telegram notifications on state changes
  • Continues until task is complete or blocked

For workflow details and phase definitions, see ~/samocode/CLAUDE.md.

Sessions: Manual vs Autonomous

There's no strict "samocode session" - just sessions. Any session can be worked on:

  • Manually by you (the parent Claude) - e.g., investigation, Q&A, planning
  • Autonomously by samocode - e.g., implementation, testing, quality fixes
  • Mixed - start manually, hand off to samocode, take back control when blocked

This flexibility is intentional. Use samocode for repetitive/long-running phases, work manually when human judgment is needed.

When to Use

Only when user explicitly asks for samocode (see Trigger Phrases above).

Do NOT assume samocode should run just because a session exists.

Execution

Arguments: $ARGUMENTS (session name or project path with session name)

Steps

  1. Find .samocode config:

    • Look for .samocode file in current working dir or project path
    • Extract the full path to the .samocode file (e.g., ~/project/.samocode)
    • If .samocode file is missing: ERROR and ask user to create it
  2. Determine session name:

    • If $ARGUMENTS is a session name (e.g., "my-task"), use it directly
    • If $ARGUMENTS includes a path, extract session name from it
    • Session will be resolved: exact match → dated match → new session
  3. Check session state (if exists):

    • Read _overview.md Status section
    • If Phase: done:
      • Ask user: "Session is complete. What new work do you want to do?"
      • Update _overview.md:
        • Phase: investigation
        • Last Action: Resuming session for: [user's goal]
        • Next: Investigate approach for [user's goal]
      • Add to Flow Log: - [MM-DD HH:MM] Resuming: [user's goal]
    • If Blocked: yes or blocked:
      • Show user the current status (Last Action, Next, reason if available)
      • Ask how to proceed
      • Update status based on user's direction
  4. Start samocode:

    bash
    cd ~/samocode && python main.py \
      --config [PATH_TO_.SAMOCODE] \
      --session [SESSION_NAME] 2>&1
    

    Optional: Add --timeout SECONDS for per-iteration time limit (default: 1800s = 30 min). Each child Claude iteration is killed if it exceeds this. Increase for complex phases:

    bash
    python main.py --config ... --session ... --timeout 3600  # 1 hour per iteration
    

    Run this in background using run_in_background: true

  5. Monitor loop using Bash + TaskOutput (CRITICAL):

    IMPORTANT: Always use TaskOutput(block=true) immediately after starting a background monitor. Do NOT rely on system notifications - they can be missed if you're mid-response or user sends a message.

    Reliable monitoring pattern:

    python
    # Step 1: Start background sleep + check
    Bash(
      command="sleep 60 && grep -E '^(Phase|Iteration|Blocked):' [SESSION]/_overview.md",
      run_in_background=true
    )
    # Returns task_id (e.g., "b155903")
    
    # Step 2: IMMEDIATELY block on result - DO NOT SKIP THIS
    TaskOutput(task_id="b155903", block=true, timeout=120000)
    # This guarantees you see the result
    
    # Step 3: Report progress to user
    
    # Step 4: If not done/blocked, repeat from Step 1
    

    Sleep duration by phase:

    • Investigation/planning: 60s (fast iterations)
    • Implementation: 120-180s (longer iterations)
    • Quality review: 120s (multi-review takes time)
    • Testing: 60s (usually quick)

    On each TaskOutput result:

    • Parse the status (Phase, Iteration, Blocked)
    • Report progress to user
    • If done or blocked → stop monitoring
    • Otherwise → start next monitor iteration (goto Step 1)

    Why this matters: Without block=true, you may lose track of monitors when:

    • User sends a message while you're waiting
    • Multiple monitors pile up
    • System notifications arrive during your response
  6. On completion or block:

    • Read final _overview.md status
    • Summarize what was accomplished
    • If blocked, explain what's needed

Required .samocode File

Every project using samocode MUST have a .samocode file in its root:

code
MAIN_REPO=~/path/to/main/repo
WORKTREES=~/path/to/worktrees/
SESSIONS=~/path/to/_sessions/

All three keys are REQUIRED:

  • MAIN_REPO: The main working directory (where Claude runs)
  • SESSIONS: Where samocode session folders are stored
  • WORKTREES: Where git worktrees are created

Session Structure

Sessions are stored in SESSIONS dir (from .samocode file), NOT nested inside projects:

code
[SESSIONS_DIR]/
└── [YY-MM-DD]-[session-name]/    # Session folder (e.g., 26-01-15-pyright-ci)
    ├── _overview.md              # Main session state
    ├── _signal.json              # Control signal
    ├── _qa.md                    # Q&A when waiting for human input
    ├── _logs/                    # Agent iteration logs (JSONL)
    │   └── [MM-DD-HHMM]-[NNN]-[phase].jsonl
    ├── [MM-DD-HH:mm]-plan-*.md   # Implementation plans
    ├── [MM-DD-HH:mm]-dive-*.md   # Investigation reports
    └── [MM-DD-HH:mm]-*.md        # Other artifacts

Key Files in _overview.md

markdown
## Status
Phase: [investigation|planning|implementation|testing|quality|done]
Iteration: N
Blocked: [yes/no]
Last Action: [what happened]
Next: [what to do next]

## Flow Log
- [NNN @ MM-DD HH:MM] Event description -> optional-file.md

Common Issues

  1. Missing .samocode file: Create .samocode file in project root with SESSIONS, WORKTREES, MAIN_REPO
  2. Telegram errors: Check ~/samocode/.env has TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID
  3. Timeout: Default is 30 min. Increase CLAUDE_TIMEOUT env var if iterations need more

Debugging Samocode Bugs

If samocode exhibits bugs or weird behavior (loops, wrong decisions, missing steps, etc.):

  1. Analyze the issue:

    • Check worker output logs for errors
    • Read _overview.md to see what went wrong
    • Check if workflow.md instructions are unclear
    • Check if skills have ambiguous or missing guidance
  2. Suggest fixes - DO NOT auto-implement:

    • Identify the root cause (workflow.md, skill, or worker code)
    • Propose specific fix to user with explanation
    • Show exact file and changes needed
    • WAIT FOR USER CONFIRMATION before making any changes
  3. Samocode source locations:

    • Worker/orchestrator: ~/samocode/main.py, ~/samocode/worker/
    • Workflow prompt: ~/samocode/workflow.md
    • Skills: ~/samocode/skills/*/SKILL.md
    • Commands: ~/samocode/commands/*.md
  4. Common fix patterns:

    • Infinite loops → Add explicit stop conditions in workflow.md
    • Wrong phase transitions → Clarify phase criteria in workflow.md
    • Missing context → Add more explicit instructions in skill
    • Format errors → Add examples in skill or workflow

IMPORTANT: Always propose fixes and wait for user approval. Samocode is critical infrastructure - no cowboy coding.

Example Usage

code
User: "Run samocode on the hvac project"
→ Find ~/code/hvac-voice-agent/.samocode file
→ Determine session name from context (e.g., "voice-agent")
→ Run: python main.py --config ~/code/hvac-voice-agent/.samocode --session voice-agent
→ Monitor iterations, report progress

User: "Continue the samocode session"
→ Find session name from context or ask user
→ Find .samocode file path
→ Run: python main.py --config [CONFIG_PATH] --session [SESSION_NAME]
→ Monitor iterations, report progress

Remember: You run python main.py, the Python worker runs Claude. You do NOT run phase agents yourself.