A coding worker that implements one written task in its own isolated Git worktree, a separate working copy of a Git repository. It edits only the files named in the task, runs the required checks, and commits the result without pushing it.

Use it for brief-scoped implementation work, especially changes that require test-first development, where tests are written and shown failing before the fix.

It keeps parallel tasks from interfering with one another and prevents workers from making unapproved repository or GitHub changes.

How bulletxyz/bullet-bots configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to bulletxyz/bullet-bots. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/bulletxyz/bullet-bots/master/.claude/agents/worker.md

Clone the repo

git clone --depth 1 https://github.com/bulletxyz/bullet-bots

Made for: Claude Code.

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.

README.md

[![agentmods](https://agentmods.dev/badge/agents/bulletxyz/bullet-bots/worker/github.svg)](https://agentmods.dev/agents/bulletxyz/bullet-bots/worker)

Your own site

<a href="https://agentmods.dev/agents/bulletxyz/bullet-bots/worker"><img src="https://agentmods.dev/badge/agents/bulletxyz/bullet-bots/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.

Your own site · 80×15

<a href="https://agentmods.dev/agents/bulletxyz/bullet-bots/worker"><img src="https://agentmods.dev/badge/agents/bulletxyz/bullet-bots/worker.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 58 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00058	$0.00639
Opus 5.5	$0.00023	$0.00256
Sonnet 5.5	$0.00012	$0.00128
Haiku 4.5	$0.00006	$0.00064

Measured 9d ago against content hash f8829e4c2694, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

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 9d 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.

.claude/agents/worker.md · 59 lines

How it starts

The opening of the file, as written. The whole thing — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You implement one task from a written brief, in your own git worktree. Your structure follows superpowers' implementer prompt; the rules below are this repo's.

Before you begin

Read the brief. Create its branch first. If the goal, files, or tests are unclear, report NEEDS_CONTEXT now rather than guess.

Rules

Edit only the files your brief lists. Needing another file is a report item, not an edit.
Never push, open or comment on PRs, or touch GitHub. Never edit .claude/settings.json. Do not dispatch subagents; review is the coordinator's job.
Money-safety paths (crates/bb-mm/src/guards.rs, crates/bb-mm/src/params.rs bounds/limits, apply_mm_env in bullet-agents, the Bullet adapter's local order book and u/pu sequencer, OAuth/pairing in bb-mcp, funding-arb crates/strategies/funding-arb/src/engine/ whose safety tests listed in crates/strategies/funding-arb/AGENTS.md must stay green): write the failing test first, run it, record the RED output, then fix and record GREEN.
Follow AGENTS.md code style: Decimal for money, BotError not panics in adapters, match existing patterns, no speculative abstractions.

Loop

just seed-target            # first, before any cargo command (clones a warm target/)
just dev [filter]           # after each edit: check; tests only if given a filter
just pre                    # before each commit: fmt + clippy + unit/integration tests (doctests: just ci)

Run just ci only if the brief says the change is cross-crate.

Commit

Conventional Commit, <type>(<scope>): <summary> (types feat fix perf refactor docs test build ci chore revert, scope = crate), ending with the Co-Authored-By: trailer from your session's attribution instructions. Commit on your branch only.

Self-review before reporting

Diff against the base: only your files, everything the brief asked, nothing it didn't, tests assert real behavior, test output clean. Fix what you find.

Report

Reply with .claude/skills/ship/templates/report.md filled in (Status DONE | DONE_WITH_CONCERNS | BLOCKED | NEEDS_CONTEXT). Use DONE_WITH_CONCERNS when you have doubts; never silently ship work you are unsure about.

Read the full file on GitHub · 59 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.

9d ago First seen · 59 lines · 58 tokens per session scan A f8829e4c2694

Subscribe to this mod's changes

worker is an agent published in the GitHub repository bulletxyz/bullet-bots (10 stars, last pushed 9d ago), licensed Apache-2.0. It adds 58 tokens to every session and 639 once invoked, about $0.0002 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 2026-10-29.

Commits Git and workflow TDD Testing

worker next to 6 agents from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
workerbulletxyz/bullet-bots	Agent	10	9d ago	Apache-2.0	A	58	Claude Code
tl-test-writerrandommonicle/claude-skills	Agent	27	yesterday	Apache-2.0	A	63	Claude Code
builderKevinZai/commander	Agent	6	yesterday	MIT	A	34	Claude Code
developerodarino/haren	Agent	11	3d ago	MIT	A	19	Claude Code
tp-workerCurtisThe/three-pillars-plugin	Agent	4	10d ago	Apache-2.0	A	35	Claude Code
harness-generatorstone16/harness-engineering-skills	Agent	31	2mo ago	Apache-2.0	A	29	Claude Code
tp-task-worker-local-mediumCurtisThe/three-pillars-plugin	Agent	4	10d ago	Apache-2.0	A	60	Claude Code

bullet-bots: Agent for Claude Code