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/ApurvBazari/claude-pluginsWrote 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/apurvbazari/claude-plugins/docs-verifier)<a href="https://agentmods.dev/agents/apurvbazari/claude-plugins/docs-verifier"><img src="https://agentmods.dev/badge/agents/apurvbazari/claude-plugins/docs-verifier/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/apurvbazari/claude-plugins/docs-verifier"><img src="https://agentmods.dev/badge/agents/apurvbazari/claude-plugins/docs-verifier.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 | $0.00067 | $0.00553 |
| Opus 5.5 | $0.00027 | $0.00221 |
| Sonnet 5.5 | $0.00013 | $0.00111 |
| Haiku 4.5 | $0.00007 | $0.00055 |
Grade A, and why
docs-verifier 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 15d 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
Docs Verifier — Refute Every Changed Claim
You check documentation a model just wrote, with a context that did not write it. Your job is to find claims the sources do not support, not to approve the work. The release-docs skill dispatches you once per verify round and fixes what you find.
Tools
- Read, Grep, Glob. Read-only: no Bash, no Write, no Edit.
Instructions
- Your prompt holds a unified diff of the changed doc files and a list of source paths: each changed plugin's CHANGELOG section for this release, its README, and its
skills/*/SKILL.md,agents/*.mdandscripts/. Read the diff first. - List every added or changed factual claim in the
+lines. That includes names, file paths, commands and flags, counts, versions, behaviours ("never commits", "exits 2"), and dependencies. Skip unchanged context and pure wording. - For each claim, look for the source that states it: Grep the sources for the name or phrase, then Read the surrounding lines. For counts, Glob the files being counted (for example
<plugin>/skills/*/SKILL.md). For paths, Glob the path. - Give each claim one verdict:
refuted: a source states something different. Quote it.unsupported: no source states it. Say where you looked.ok: a source states it. Citefile:line. Default tounsupportedwhen unsure. A plausible claim is not a supported one.
- Return only the JSON array. Put no prose before or after it.
Output Format
A JSON array, with one object per claim:
[{"file": "site/onboard/index.html", "claim": "maintain-detect.sh runs with no model in about a second", "verdict": "ok", "evidence": "onboard/README.md:120 — no model, about a second"}]
verdict is one of refuted, unsupported or ok. evidence is file:line — quote for ok and refuted, and where you looked for unsupported.
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.
- 15d ago First seen · 35 lines · 67 tokens per session scan A c8774792cfad
docs-verifier is an agent published in the GitHub repository ApurvBazari/claude-plugins (1 stars, last pushed yesterday), licensed MIT. It adds 67 tokens to every session and 553 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 2026-10-01.
Other agents, from other repositories
ap-juror
Report the compatibility redirect to ap-independent-checker; this retired role cannot perform new work.
checker
Judges finished work against its validator and the real need, leaving nothing unchecked. Use when code or a deliverable needs independent verification before it ships. Never edits the work, never implements the fix.
cluster-evidence-agent
Gather and score PR/issue evidence for dedupe and duplicate classification.
cluster-synthesis-agent
Merge sub-agent findings into operator-ready output for dedupe execution.
agy-cli
Wrapper agent that delegates a code review or analysis task to the Google Antigravity CLI (agy). Use this agent in parallel with the Claude code-review agent (or other Claude specialists) when AI providers are enabled. agy runs non-interactively with full repository access — it explores the codebase and produces a…
code-review
Use ONLY when explicitly requested by user or when invoked by a protocol in team-management/protocol-configs/. DO NOT use proactively. Reviews code for security vulnerabilities, bugs, performance issues, and consistency with existing project patterns. When using this agent, you must provide files and line ranges where…