agent-worker

agent-worker is a skill for Claude Code, Codex from 02loveslollipop/OpenCROW. It costs 35 tokens per session (639 once invoked), scanned A, original, Apache-2.0.

A task-delegation tool for sending implementation work to selected local coding workers and tracking their progress. It can use separate Git worktrees, which are isolated working folders for branches.

In plain words
What is it for?
Use it to assign a concrete coding task with acceptance criteria, inspect whether it passed, answer worker questions, and continue from checkpoints.
Why use it?
It removes the need to manually coordinate separate coding processes and lose track of their status, questions, results, or handoffs.

Skill for Claude CodeCodex

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

Good fit Use it to assign a concrete coding task with acceptance criteria, inspect whether it passed, answer worker questions, and continue from checkpoints.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/02loveslollipop/opencrow/agent-worker
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add 02loveslollipop/OpenCROW --skill agent-worker
Clone the repo
git clone --depth 1 https://github.com/02loveslollipop/OpenCROW

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 agent-worker

README.md
[![agentmods](https://agentmods.dev/badge/skills/02loveslollipop/opencrow/agent-worker/github.svg)](https://agentmods.dev/skills/02loveslollipop/opencrow/agent-worker)
Your own site
<a href="https://agentmods.dev/skills/02loveslollipop/opencrow/agent-worker"><img src="https://agentmods.dev/badge/skills/02loveslollipop/opencrow/agent-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 agent-worker

Your own site · 80×15
<a href="https://agentmods.dev/skills/02loveslollipop/opencrow/agent-worker"><img src="https://agentmods.dev/badge/skills/02loveslollipop/opencrow/agent-worker.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 35 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 639 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 $0.00035 $0.00639
Opus 5.5 $0.00014 $0.00256
Sonnet 5.5 $0.00007 $0.00128
Haiku 4.5 $0.00003 $0.00064

Measured 1mo ago against content hash bc3147286c03, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

agent-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 1mo 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/agent-worker/SKILL.md · 27 lines

How it starts

The opening of the file, as written. The whole thing — 27 lines — stays where its author put it; the contents beside it link to each section on GitHub.

OpenCROW Runner - Agent Worker

Use opencrow-worker-mcp to delegate a concrete task to an explicitly selected provider. The standalone runner requires a full OpenCROW installation and the chosen authenticated CLI; it does not require Constellation or an SDK. Inspect toolbox_verify when checking provider availability. Model availability and credentials are verified only when a turn runs.

Delegate and supervise

  • Call worker_start with task, provider (antigravity, opencode, or codex), and workspace. Include acceptance criteria and an optional provider-native model and effort. Workers execute with trusted full host access; delegation preserves the scope of the user's task.
  • Start returns a logical worker_id before the task completes. Its successful MCP envelope means the request was recorded, not that the task passed. Inspect the worker's state and results.
  • Default workspace_mode: auto creates a unique Git worktree and branch from HEAD, excluding uncommitted source edits. Outside Git, it uses the supplied directory. Use shared explicitly to operate on the same files and checked-out branch, including local edits. Concurrent shared workers can overwrite each other; assign distinct responsibilities.
  • Read worker_events with the last returned next_cursor. Use optional wait_sec up to 30 seconds to wait for updates, or omit worker_id to monitor all workers. Use worker_status for current state and artifact paths.
  • Answer a question event using worker_reply with its question_id. Replies are durable; an early reply waits for the current turn to finish. Workers ask cooperatively and end their turn; this is not native interactive steering.
  • Use worker_followup after completion to refine work in the same native session. Select an explicit model override when needed. Pending questions use worker_reply instead.
  • Review the retained worktree, diff, checkpoint, final response, and test evidence. Integrate and clean up through the normal repository workflow; the runner does not auto-commit, merge, push, or remove files.

Read the full file on GitHub · 27 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 1mo ago First seen · 27 lines · 35 tokens per session scan A bc3147286c03

Subscribe to this mod's changes

agent-worker is a skill published in the GitHub repository 02loveslollipop/OpenCROW (10 stars, last pushed today), licensed Apache-2.0. It adds 35 tokens to every session and 639 once invoked, about $0.0001 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 2026-09-08.