技能 编程开发 委派子代理完成任务

委派子代理完成任务

v20260803
sub-agent
本技能提供了一个通用的入口点,用于将复杂的、自包含的任务委派给各种大型语言模型(如Codex、Claude Code、Pi)驱动的子代理。它适用于元技能流程图(DAG)中的步骤,执行功能创建、PR审查、代码重构或文档生成等任务。建议使用非交互式CLI模式进行可靠的后台执行。
获取技能
455 次下载
概览

Sub-Agent (opensquilla process tools)

Generic "spawn a sub-Agent" entry point for delegating self-contained tasks to Codex / Claude Code / OpenCode / Pi via background process. Wrapping CLIs are coding-oriented, but the skill itself is used as the generic sub-Agent slot by meta-skill DAGs for any LLM-driven sub-task (file edits, document generation, policy review, etc.).

Use opensquilla's exec_command, background_process, and process tools for coding agent work. OpenSquilla does not expose a bash tool; do not use the legacy bash tool-call DSL.

Non-Interactive CLI Mode

OpenSquilla's process tools do not expose a pty parameter. Prefer non-interactive command modes that run and exit cleanly:

# ✅ Correct for Codex/Pi/OpenCode
exec_command(command="codex exec 'Your prompt'")

For Claude Code (claude CLI), use --print --permission-mode bypassPermissions instead. --dangerously-skip-permissions with PTY can exit after the confirmation dialog. --print mode keeps full tool access and avoids interactive confirmation:

# ✅ Correct for Claude Code (no PTY needed)
cd /path/to/project && claude --permission-mode bypassPermissions --print 'Your task'

# For background execution: use background_process

# ❌ Wrong for Claude Code
exec_command(command="claude --dangerously-skip-permissions 'task'")

OpenSquilla Tool Parameters

Tool Key parameters Description
exec_command command, workdir, timeout Run a foreground shell command.
background_process command, workdir, timeout Start a long-running command and return session_id.
process action, session_id, data, offset, limit Poll, log, write to, or stop a background process.

Process Tool Actions (for background sessions)

Action Description
list List all running/recent sessions
poll Check if session is still running
log Get session output (with optional offset/limit)
write Send raw data to stdin
submit Send data + newline (like typing and pressing Enter)
eof Close stdin
remove Remove a finished session from the process list
kill Terminate the session

Quick Start: One-Shot Tasks

For quick prompts/chats, create a temp git repo and run:

# Quick chat (Codex needs a git repo!)
SCRATCH=$(mktemp -d) && cd $SCRATCH && git init && codex exec "Your prompt here"

# Or in a real project
exec_command(workdir="~/Projects/myproject", command="codex exec 'Add error handling to the API calls'")

Why git init? Codex refuses to run outside a trusted git directory. Creating a temp repo solves this for scratch work.


The Pattern: workdir + background_process

For longer tasks, use background_process:

# Start agent in target directory.
background_process(workdir="~/project", command="codex exec --full-auto 'Build a snake game'")
# Returns session_id for tracking

# Wait for it to finish — blocks until the process exits (or the timeout
# elapses, in which case just call wait again). Prefer this over polling in a
# loop: a looped process(action="poll") burns a full turn + tokens each time.
process(action="wait", session_id="XXX")

# Peek at output without blocking (optional, for progress)
process(action="log", session_id="XXX")

# Send input (if agent asks a question)
process(action="write", session_id="XXX", data="y")

# Submit with Enter (like typing "yes" and pressing Enter)
process(action="submit", session_id="XXX", data="yes")

# Kill if needed
process(action="kill", session_id="XXX")

Why workdir matters: Agent wakes up in a focused directory, doesn't wander off reading unrelated files (like your soul.md 😅).


Codex CLI

Model: gpt-5.2-codex is the default (set in ~/.codex/config.toml)

Flags

Flag Effect
exec "prompt" One-shot execution, exits when done
--full-auto Sandboxed but auto-approves in workspace
--yolo NO sandbox, NO approvals (fastest, most dangerous)

Building/Creating

# Quick one-shot
exec_command(workdir="~/project", command="codex exec --full-auto 'Build a dark mode toggle'")

# Background for longer work
background_process(workdir="~/project", command="codex exec --full-auto 'Refactor the auth module'")

Reviewing PRs

⚠️ CRITICAL: Never review PRs in OpenSquilla's own project folder! Clone to temp folder or use git worktree.

# Clone to temp for safe review
REVIEW_DIR=$(mktemp -d)
git clone https://github.com/user/repo.git $REVIEW_DIR
cd $REVIEW_DIR && gh pr checkout 130
exec_command(workdir="$REVIEW_DIR", command="codex review --base origin/main")
# Clean up after: trash $REVIEW_DIR

# Or use git worktree (keeps main intact)
git worktree add /tmp/pr-130-review pr-130-branch
exec_command(workdir="/tmp/pr-130-review", command="codex review --base main")

Batch PR Reviews (parallel army!)

# Fetch all PR refs first
git fetch origin '+refs/pull/*/head:refs/remotes/origin/pr/*'

# Deploy the army - one Codex per PR
background_process(workdir="~/project", command="codex exec 'Review PR #86. git diff origin/main...origin/pr/86'")
background_process(workdir="~/project", command="codex exec 'Review PR #87. git diff origin/main...origin/pr/87'")

# Monitor all
process(action="list")

# Post results to GitHub
gh pr comment <PR#> --body "<review content>"

Claude Code

# Foreground
exec_command(workdir="~/project", command="claude --permission-mode bypassPermissions --print 'Your task'")

# Background
background_process(workdir="~/project", command="claude --permission-mode bypassPermissions --print 'Your task'")

OpenCode

exec_command(workdir="~/project", command="opencode run 'Your task'")

Pi Coding Agent

# Install: npm install -g @mariozechner/pi-coding-agent
exec_command(workdir="~/project", command="pi -p 'Your task'")

# Non-interactive mode
exec_command(command="pi -p 'Summarize src/'")

# Different provider/model
exec_command(command="pi --provider openai --model gpt-4o-mini -p 'Your task'")

Note: Pi now has Anthropic prompt caching enabled (PR #584, merged Jan 2026)!


Parallel Issue Fixing with git worktrees

For fixing multiple issues in parallel, use git worktrees:

# 1. Create worktrees for each issue
git worktree add -b fix/issue-78 /tmp/issue-78 main
git worktree add -b fix/issue-99 /tmp/issue-99 main

# 2. Launch Codex in each
background_process(workdir="/tmp/issue-78", command="pnpm install && codex exec --full-auto 'Fix issue #78: <description>. Commit and push.'")
background_process(workdir="/tmp/issue-99", command="pnpm install && codex exec --full-auto 'Fix issue #99 from the approved ticket summary. Implement only the in-scope edits and commit after review.'")

# 3. Monitor progress
process(action="list")
process(action="log", session_id="XXX")

# 4. Create PRs after fixes
cd /tmp/issue-78 && git push -u origin fix/issue-78
gh pr create --repo user/repo --head fix/issue-78 --title "fix: ..." --body "..."

# 5. Cleanup
git worktree remove /tmp/issue-78
git worktree remove /tmp/issue-99

⚠️ Rules

  1. Use the right execution mode per agent:
    • Codex/Pi/OpenCode: non-interactive command mode (codex exec, opencode run, pi -p)
    • Claude Code: --print --permission-mode bypassPermissions (no PTY required)
  2. Respect tool choice - if user asks for Codex, use Codex.
    • Orchestrator mode: do NOT hand-code patches yourself.
    • If an agent fails/hangs, respawn it or ask the user for direction, but don't silently take over.
  3. Be patient - don't kill sessions because they're "slow"
  4. Monitor with process:log - check progress without interfering
  5. --full-auto for building - auto-approves changes
  6. vanilla for reviewing - no special flags needed
  7. Parallel is OK - run many Codex processes at once for batch work
  8. NEVER start Codex inside your OpenSquilla state directory ($OPENSQUILLA_STATE_DIR, default ~/.opensquilla/state) - keep agent state separate from project worktrees.
  9. NEVER checkout branches inside the live OpenSquilla runtime state/workspace directories - use an explicit project worktree.

Progress Updates (Critical)

When you spawn coding agents in the background, keep the user in the loop.

  • Send 1 short message when you start (what's running + where).
  • Then only update again when something changes:
    • a milestone completes (build finished, tests passed)
    • the agent asks a question / needs input
    • you hit an error or need user action
    • the agent finishes (include what changed + where)
  • If you kill a session, immediately say you killed it and why.

This prevents the user from seeing only "Agent failed before reply" and having no idea what happened.


Auto-Notify on Completion

For long-running background tasks, ask the agent to print a clear completion line so progress is visible in process(action="log", ...) output:

... your task here.

When completely finished, send a brief status update in this session.

Example:

background_process(workdir="~/project", command="codex exec --full-auto 'Build a REST API for todos.

When completely finished, print: Done: Built todos REST API with CRUD endpoints'")

This makes completion visible in the background process log.


Learnings (Jan 2026)

  • Prefer non-interactive modes: Coding agents are easiest to supervise when they print progress and exit cleanly.
  • Git repo required: Codex won't run outside a git directory. Use mktemp -d && git init for scratch work.
  • exec is your friend: codex exec "prompt" runs and exits cleanly - perfect for one-shots.
  • submit vs write: Use submit to send input + Enter, write for raw data without newline.
  • Sass works: Codex responds well to playful prompts. Asked it to write a haiku about being second fiddle to a space lobster, got: "Second chair, I code / Space lobster sets the tempo / Keys glow, I follow" 🦞
信息
Category 编程开发
Name sub-agent
版本 v20260803
大小 13.18KB
更新时间 2026-08-04
语言