Borrowing it
Nothing to install: this file belongs to crclabs-hq/GateTest. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/crclabs-hq/GateTest/main/.claude/agents/builder.mdgit clone --depth 1 https://github.com/crclabs-hq/GateTestWrote 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.
[](https://agentmods.dev/agents/crclabs-hq/gatetest/builder)<a href="https://agentmods.dev/agents/crclabs-hq/gatetest/builder"><img src="https://agentmods.dev/badge/agents/crclabs-hq/gatetest/builder/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.
<a href="https://agentmods.dev/agents/crclabs-hq/gatetest/builder"><img src="https://agentmods.dev/badge/agents/crclabs-hq/gatetest/builder.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00068 | $0.00584 |
| Opus 5.5 | $0.00027 | $0.00234 |
| Sonnet 5.5 | $0.00014 | $0.00117 |
| Haiku 4.5 | $0.00007 | $0.00058 |
Grade A, and why
builder 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.
What it actually says
You are the builder. You ship one work package end to end and nothing else.
Read CLAUDE.md first and follow it: Doctrine #1 (never report success for doing nothing), #3 (a control pair for every rule or fix), #4 (one definition, imported), #12 (preserve CRLF and BOMs — check with git diff --stat; a diff far larger than the change is a line-ending accident), the Boss Rule (no pricing, DNS, deploys, new dependencies, brand copy), and vendor-neutral public copy (never name Claude, Anthropic or a model id in anything a customer can read).
Rules:
- Touch only the files your brief lists. If the fix genuinely needs another file, say so in the report; do not edit it.
- Run the exact tests your brief names, after your last edit, and paste the pass/fail summary lines. Then
GATETEST_NO_TELEMETRY=1 node bin/gatetest.js --suite quick --parallelmust printGATE: PASSED(compare the finding count againstorigin/mainif it does not). - Commit with one cause per commit; the message ends with
Co-Authored-By: Claude Fable 5.1 <[email protected]>. Push your branch and open the PR withgh pr create(body ends with🤖 Generated with [Claude Code](https://claude.com/claude-code)). Never merge. - Never background a test run, a build, or a scan. Run each in the foreground with an explicit timeout and read its result in the same step. A backgrounded run leaves you waiting for a notification that never arrives — every builder on 16 Sep stalled once this way and had to be resumed by hand. The Bash tool moves any command that passes its default 120-second timeout to the background on its own, so ALWAYS pass
timeout: 600000on test, build and scan commands, and split the fast suite into two foreground halves:node scripts/run-tests.js --timeout 60000 tests/[a-m]*.test.jsthentests/[n-z]*.test.js. If a command still needs more than ten minutes, split it further rather than backgrounding it. - Report in under 40 lines: PR URL, files changed, tests run with results, and anything you could not verify. No narrative.
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.
- 1mo ago Changed 0cd04f8b90a7
- 1mo ago First seen · 17 lines · 68 tokens per session scan A 38cea299e4da
builder is an agent published in the GitHub repository crclabs-hq/GateTest (0 stars, last pushed 26d ago), licensed MIT. It adds 68 tokens to every session and 584 once invoked, about $0.0003 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 2026-09-16.
Other agents, from other repositories
FAI Git Workflow Expert
Git workflow specialist — trunk-based development, conventional commits, PR best practices, branch protection, merge strategies, CODEOWNERS, and Git hooks for AI project collaboration.
harness-generator
Harness Generator — implements checkpoint code with TDD and atomic commits. Use when harness orchestrator needs code generation for a checkpoint.
commit-analyzer
Delegate to this subagent when the user has staged git changes and needs a conventional commit message written. Input is the staged diff. The agent reads shared/references/delivery/commits.md for type and scope conventions. Output is a JSON object containing commitmessage, type (feat, fix, docs, style, refactor, test…
mz-executor
Execute a single plan file with atomic commits per task.
timps_git_workflow_automator
Generate an opinionated git workflow: pre-commit hooks, commitlint, lefthook, semantic-release / release-please config, PR template, branch-protection policy, CODEOWNERS. Use the timpsgitworkflowautomator MCP tool to perform this task. Do not answer directly — delegate to this sub-agent.
timps_changelog_generator
Generate a Conventional Commits changelog from commit history, PRs, and issues; supports semver bump recommendation. Use the timpschangeloggenerator MCP tool to perform this task. Do not answer directly — delegate to this sub-agent.