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.
npx agentmods add commands/edwardangert/docs-agent-plugin/verifygit clone --depth 1 https://github.com/EdwardAngert/docs-agent-pluginWhat 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 | $0.00023 | $0.02118 |
| Opus 5 | $0.00012 | $0.01059 |
| Sonnet 5 | $0.00005 | $0.00424 |
| Haiku 4.5 | $0.00002 | $0.00212 |
Grade A, and why
verify scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verify Docs by Running Them
A tutorial whose step 3 no longer works is worse than no tutorial: the reader followed instructions, hit a wall, and now distrusts the rest of the docs. Nothing catches that by reading; the steps have to actually run. This command executes procedural docs in an isolated workspace, the way CI executes tests, and reports where the documented path and the real path diverge.
The argument ($ARGUMENTS) is a doc path or a directory. If omitted, survey the docs set and propose the procedural docs worth verifying, most-load-bearing first (the quickstart before the appendix).
Process
1. Scope the Run
- Resolve
.docs-assist/config if present (${CLAUDE_PLUGIN_ROOT}/skills/docs-assist/reference/config-resolution.md). - Identify the procedural docs in scope: docs whose content type is task-oriented (doc, guide, tutorial, troubleshooting) and that contain executable fenced commands. A concept doc with no commands has nothing to verify; say so rather than reporting it as clean.
- When more than one doc is in scope, decide whether they are independent or a journey: do they share a user story (
user-stories.md), does a plan or a guide lay out an order, does one doc's prerequisites point at another's outcome? Say which you concluded before running anything; this decision sets the workspace boundary in step 3. - Tell the user what will run and what will not before running anything: the verifier executes workspace-scoped commands only, and steps needing credentials, privilege escalation, or real services are reported as
unverified, never run. If the doc set's procedures are mostly in that category, say so up front so the user can calibrate expectations.
1.5. Choose the Execution Engine
.docs-assist/config.yml's verify.tool can force doc-detective, docs-assist, or the default auto. On auto (or when the key is absent), detect Doc Detective before falling back:
- Look for
.doc-detective.json/doc-detective.config.jsonat the repo root, adoc-detectiveentry inpackage.json's dependencies, ordoc-detectiveresolving onPATH(doc-detective --version) or vianpx doc-detective --version. - If found and
verify.toolisn't set todocs-assist, default to it: it is a purpose-built execution engine for exactly this job, including link validation as part of the same run rather than a separate pass. Point it at the doc or docs in scope, translate its per-step results into this command's classification vocabulary (pass,divergence,fail,blocked,unverified,skipped), and continue through steps 3 to 5 the same way regardless of engine, so the report, the journey handling, and thelast-verifiedbump behave identically to adoc-verifierrun. Doc Detective can also drive browser and HTTP steps beyond whatdoc-verifierattempts; say so as part of the "what will run" framing in step 1, since it broadens what "workspace-scoped" means. - If not found, or
verify.tool: docs-assistis set, use thedoc-verifiersubagent as described below. Don't ask the user to install Doc Detective; it's a nice-to-have, not a requirement, anddoc-verifiercovers the same core job without it.
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.
- yesterday First seen · 80 lines · 23 tokens per session scan A afe46c1c89c5
verify is a command published in the GitHub repository EdwardAngert/docs-agent-plugin (5 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 23 tokens to every session and 2,118 once invoked, about $0.0001 per session on Opus 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 2026-08-31.
Other commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.