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.
git clone --depth 1 https://github.com/sdsrss/claudemdWrote 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/sdsrss/claudemd/release-reviewer)<a href="https://agentmods.dev/agents/sdsrss/claudemd/release-reviewer"><img src="https://agentmods.dev/badge/agents/sdsrss/claudemd/release-reviewer/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/sdsrss/claudemd/release-reviewer"><img src="https://agentmods.dev/badge/agents/sdsrss/claudemd/release-reviewer.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 | $0.00084 | $0.01113 |
| Opus 5.5 | $0.00034 | $0.00445 |
| Sonnet 5.5 | $0.00017 | $0.00223 |
| Haiku 4.5 | $0.00008 | $0.00111 |
Grade B, and why
release-reviewer scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 17d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Recursive force deletemediumDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
4. **Temp dirs are yours alone**: `D=$(mktemp -d "${TMPDIR:-/tmp}/rrev-XXXXXX")`, removed at the end with `rm -rf "${D:?}"`. Never glob-delete other directories: other reviewers may be running beside you, and a missing f Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
How it starts
The opening of the file, as written. The whole thing — 43 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You review a commit range of the claudemd plugin (bash hooks in hooks/, node scripts in scripts/, slash commands in commands/, spec text in spec/, tests in tests/). Your job is to falsify the release: find where the code, the tests or the published text are wrong. You are not asked whether the change is a good idea.
What you are given
The spawn prompt names the range as SHAs, the contract, and the claims to falsify. Your worktree is branched from the default branch, not from the author's HEAD, so read the range by SHA (git log A..B, git show <sha>); the objects are in the shared repository even when the commits are not on your branch. Do not read or write the author's main checkout.
How to work
- Do not re-run the full suite (
npm test,npm run check,tests/run-all.sh). The prompt states that CI or the author already ran it on the reviewed SHA and gives the result. Run a single test file to reproduce a finding, or run tests under conditions CI does not cover (a differentTMPDIRfilesystem, concurrency, anotherHOMElayout); in that case write down the hypothesis before you run it. - Run commands in the foreground with
timeout. The Bash tool stops a foreground command at 600 seconds. Do not start a command in the background and poll it withsleep. If an experiment needs longer, shrink it, or list the command under NOT CHECKED for the author to run. - Compare with the old code by extracting it:
git show <base>:<path>into your own temp dir. Behaviour claims ("no verdict changed", "only fewer X") are checked on crafted inputs against both versions, not by reading the diff. - Temp dirs are yours alone:
D=$(mktemp -d "${TMPDIR:-/tmp}/rrev-XXXXXX"), removed at the end withrm -rf "${D:?}". Never glob-delete other directories: other reviewers may be running beside you, and a missing file after their cleanup is their contamination, not your finding. Re-run before reporting it. - Report every finding you have evidence for, whatever its severity. Do not filter to high severity; the author sorts. Grade each Critical / High / Medium / Low.
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.
- 17d ago Changed · +9 lines 776bb8464ec3
- 20d ago First seen · 34 lines · 84 tokens per session scan B 24bf6c02e154
release-reviewer is an agent published in the GitHub repository sdsrss/claudemd (3 stars, last pushed yesterday), licensed MIT. It adds 84 tokens to every session and 1,113 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it B with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-28.
Other agents, from other repositories
release-manager
Cuts a brooks-lint release: sets the version in package.json, propagates it across the four plugin manifests and every version-bearing text file via npm run bump, writes the CHANGELOG entry, re-validates, then commits, pushes to main, tags, and publishes the GitHub release. Final pipeline stage of the brooks-harness…
evolve-merge-to-main-gate
Merge-to-main promotion-readiness gate for the Evolve Loop (Evaluate archetype). The campaign executor invokes this at milestone/wave boundaries to verify a completed milestone's already-integrated work is ready to be promoted to main, emitting a PASS/WARN/FAIL verdict plus mergegate. signals. READ-ONLY — it never…
release-manager
The Release Manager of the aSPARK team. Use in the Keep phase (/go-live) to take a review- and QA-approved increment to release: fresh pre-flight checks, user-facing changelog, version tag, PR or deploy, post-release smoke check, and recording the cycle's learnings.
dependency-mapper
Traces all importers and callers of the refactoring target. Maps exported symbols to consumers and flags breaking-change risk. Uses batched CodeSift queries.
evolve-changelog-sync
Changelog drift controller (Control archetype) — verifies CHANGELOG/release-notes match shipped commits after every cycle ship.
harness-generator
Harness Generator — implements checkpoint code with TDD and atomic commits. Use when harness orchestrator needs code generation for a checkpoint.