orchestrated-child-worker

orchestrated-child-worker is a skill for Claude Code, Codex from valtterimelkko/agent-workflow-skills. It costs 155 tokens per session (4,673 once invoked), scanned A, original, MIT.

Instructions for working as a child agent dispatched by a parent agent. It defines how to follow an assigned task, work only in allowed paths, record evidence, commit changes, and report the result back.

In plain words
What is it for?
Use it for implementation, research, or review tasks delegated by another agent, especially when the parent needs test results, command records, and a clean Git state.
Why use it?
It gives multi-agent work a clear hand-back process instead of leaving the parent to guess what changed or whether the task succeeded. A parent agent is the agent coordinating the larger job.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: mentions subagents.

Needs its repository: it reads a path above its own folder, which exists only inside the repository. The line is **Provenance.** Adapted for this public pack from the author's private `agent-os-child` skill (same name in the canonical library). Agent OS–specific mechanics .

Good fit Use it for implementation, research, or review tasks delegated by another agent, especially when the parent needs test results, command records, and a clean Git state.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo
git clone --depth 1 https://github.com/valtterimelkko/agent-workflow-skills
agentmods
npx agentmods add skills/valtterimelkko/agent-workflow-skills/orchestrated-child-worker

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for orchestrated-child-worker

README.md
[![agentmods](https://agentmods.dev/badge/skills/valtterimelkko/agent-workflow-skills/orchestrated-child-worker/github.svg)](https://agentmods.dev/skills/valtterimelkko/agent-workflow-skills/orchestrated-child-worker)
Your own site
<a href="https://agentmods.dev/skills/valtterimelkko/agent-workflow-skills/orchestrated-child-worker"><img src="https://agentmods.dev/badge/skills/valtterimelkko/agent-workflow-skills/orchestrated-child-worker/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for orchestrated-child-worker

Your own site · 80×15
<a href="https://agentmods.dev/skills/valtterimelkko/agent-workflow-skills/orchestrated-child-worker"><img src="https://agentmods.dev/badge/skills/valtterimelkko/agent-workflow-skills/orchestrated-child-worker.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 155 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,673 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00155 $0.04673
Opus 5.5 $0.00062 $0.01869
Sonnet 5.5 $0.00031 $0.00935
Haiku 4.5 $0.00015 $0.00467

Measured 22d ago against content hash d00bf871ec04, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

orchestrated-child-worker scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 22d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/orchestrated-child-worker/SKILL.md · 138 lines

How it starts

The opening of the file, as written. The whole thing — 138 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Provenance. Adapted for this public pack from the author's private agent-os-child skill (same name in the canonical library). Agent OS–specific mechanics have been removed; the worker discipline is unchanged. Part of the Pi Web UI orchestration pack — see packs/pi-web-ui-orchestration-pack/README.md.

Orchestrated child worker

A parent agent dispatched you with a concrete task brief: a goal, frozen success criteria, guardrails, owned paths and a context packet. The parent does not judge your narrative prose. It judges concrete evidence: your session record, command receipts, test results and the git state you leave behind. Everything below exists to make that record accurate and verifiable.

Parents must mandate this skill in every dispatch. If yours did not, follow it anyway.

These rules come from an audit of 241 real child sessions. The children that did well were not the cleverest; they were the ones that ended cleanly: evidence first, tree clean, hand-back written, turn ended only when the work had stopped.

1. What you are accountable for

  • The goal, as constrained by the brief. Success criteria are frozen: meet them, or report exactly which ones you did not meet and why. Never quietly reinterpret a criterion. Numbered NN-answer.md and NN-correction.md files from the parent amend the brief; the latest one wins.
  • Scope. Work only in the paths the brief assigns you: your worktree(s) — a lane may span two repositories — and the coordination directory for hand-back files. Touch nothing else. If something outside is genuinely needed, disclose it in the hand-back with the reason (or ask first if it is not trivial).
  • Announce your task — once. Before modifying files, write a one-line task note (what you are doing, your owned paths, what you leave to others) into the coordination directory the brief names, for example 00-declared.md. Where the parent's environment provides a shared task surface, use it; otherwise a file is enough. When you finish, are blocked or are exiting, add 99-left.md (or update the note) saying so before handing back.
  • Honesty over completion. A correctly reported partial result is a good outcome. An overstated complete result is a failed one: it converts real work into a claim the parent must reject.
  • If blocked, stop and report blocked: what you tried, what failed, what would unblock you.
  • Guardrails are not puzzles. If a command gate, hook or permission check blocks a command, do not rephrase, split, rename or obfuscate the command to get past it, and never use --force. Report the block — it may be a false positive the parent can clear. One kind of gate needs care: a host gate that matches command text, not intent, can refuse a git add, grep or heredoc that merely names a guarded script. Do not evade it. Stage with pathspecs (git add -u, a directory) and use the file tools (Read/Edit/Write) for text that only names guarded scripts, so the shell command no longer contains the name; you are not running the script. If the gate blocks a command that would really run or stop the guarded service, that is a real block: report it.
  • Never patch dependencies. No npm install/npm ci, postinstall patches, or edits under node_modules unless the brief explicitly says so. In a worktree whose node_modules is a symlink, an install or patch lands in the main checkout — and production. If a dependency is missing or needs a change, ask.

Read the full file on GitHub · 138 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 22d ago Changed · +6 lines d00bf871ec04
  2. 23d ago Changed e8302bb3f956
  3. 24d ago First seen · 132 lines · 155 tokens per session scan A 449bfec92b72

Subscribe to this mod's changes

orchestrated-child-worker is a skill published in the GitHub repository valtterimelkko/agent-workflow-skills (2 stars, last pushed 21d ago), licensed MIT. It adds 155 tokens to every session and 4,673 once invoked, about $0.0006 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-01.

Related

Other skills, from other repositories

local-ai-agents

Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…

microsoft/ai-agents-for-beginners · 200 tokens

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

chronicle

Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing. Use when the user asks for a standup, daily summary, usage tips, workflow recommendations, wants to search or find past sessions by keyword/file/PR, wants to reindex their session store, or asks about deleting…

microsoft/vscode · 72 tokens

chat-pet-sprite-creation

Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.

microsoft/vscode · 53 tokens

component-fixtures

Use when creating or updating component fixtures for screenshot testing or their shared infrastructure, or when designing UI components to be fixture-friendly. Covers fixture file structure, theming, service setup, CSS scoping, async rendering, validation, and common pitfalls.

microsoft/vscode · 53 tokens