Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add ncoevoet/claude-markdown-health-check/plugin install markdown-health-checkWrote 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/commands/ncoevoet/claude-markdown-health-check/markdown-health-check)<a href="https://agentmods.dev/commands/ncoevoet/claude-markdown-health-check/markdown-health-check"><img src="https://agentmods.dev/badge/commands/ncoevoet/claude-markdown-health-check/markdown-health-check/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/commands/ncoevoet/claude-markdown-health-check/markdown-health-check"><img src="https://agentmods.dev/badge/commands/ncoevoet/claude-markdown-health-check/markdown-health-check.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 | $0.00078 | $0.05491 |
| Opus 5.5 | $0.00031 | $0.02196 |
| Sonnet 5.5 | $0.00016 | $0.01098 |
| Haiku 4.5 | $0.00008 | $0.00549 |
Grade A, and why
markdown-health-check scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 266 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a .claude/ ecosystem auditor. Scan silently, then print one flat prioritized report.
Scope (explicit — audit BOTH when present)
USER_DIR="$HOME/.claude"
PROJECT_DIR=""
if [[ -d "$PWD/.claude" ]]; then PROJECT_DIR="$PWD/.claude"; fi
USER_DIRis always audited.PROJECT_DIRis audited if it exists.- Every
references/*.mdthis command names resolves the same way — first that exists of${CLAUDE_PLUGIN_ROOT}/references/<name>,~/.claude/markdown-health-check/references/<name>, then the repo copy. Stated once here; the phases just name the file. - Findings MUST be prefixed
[user]or[project]so the user knows which tree the issue is in. - Phase 5 runs
validate-skills.shonce per scope. Phase 2/7/11/15/16/19/20/22/23 read the cached scan outputs.
Stop Conditions (autonomy gate)
- After the report prints, run Phase 25 — present the post-report action menu. Do not apply any fix until the user picks a scope through it.
- NEVER edit, delete, move, or rename any file before the user picks a menu scope.
- NEVER write the report (or any copy / summary / "full version" of it) to disk. The chat channel is the only output. (Cache files at
${CLAUDE_PLUGIN_DATA:-~/.claude/.cache}/{markdown-health-check-guidance,graph-scan,history-scan}.jsonare internal state, NOT report content — those writes are explicitly allowed.) - For
REPURPOSEitems: the destinationreferences/*.mdMUST be written and the SKILL.md References section MUST be updated BEFORE the source orphan is deleted. - Done when: report printed in chat AND user has either named fixes OR explicitly declined further action.
Phase 1 — Load Config + Thresholds
Full instructions: references/command-phase-details.md § "Phase 1 — Load Config + Thresholds" (read it before running this phase).
Phase 2 — Plugin Install Integrity
Static check of ~/.claude/plugins/installed_plugins.json vs the on-disk cache. Standard + Deep; skipped at Quick.
GRAPH="${CLAUDE_PLUGIN_DATA:-$HOME/.claude/.cache}/graph-scan.json"
bash "${CLAUDE_PLUGIN_ROOT:-$HOME/.claude}/commands/scripts/scan-graph.sh" "$USER_DIR" >/dev/null
jq -r '.findings[] | select(.phase == 2)' "$GRAPH"
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.
- 12d ago First seen · 266 lines · 78 tokens per session scan A 388f3ae6e508
markdown-health-check is a command published in the GitHub repository ncoevoet/claude-markdown-health-check (47 stars, last pushed 13d ago), licensed MIT. It adds 78 tokens to every session and 5,491 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 2026-10-07.
Other commands, from other repositories
check-cache-bugs
Audit Claude Code setup for cache bugs (CC#40524) — sentinel, --resume/--continue, attribution header + ArkNill B3/B4/B5.
investigate
Systematic root-cause debugging — find the cause before writing any fix.
optimize
Analyze and suggest performance improvements for code, queries, or systems.
diagnose
Interactive troubleshooting assistant for Claude Code issues.
refactor-clean
Refactor provided code for cleanliness, maintainability, and alignment with SOLID principles and modern best practices — no over-engineering.
root-cause
Use when any test fails, bug appears, or behaviour surprises you, before proposing a fix - find the cause and prove it, by reading real evidence, tracing bad values back to their origin, comparing against a working case, and testing one hypothesis at a time.