What is check-on-claude?

A self-check command named Maude that reviews Claude’s recent work for repeated tool calls, unread instructions, stale project information, and unsupported claims.

Use it to inspect an assistant’s trace, find repeated searches, check whether project instructions were read, and identify possible confabulation.

It helps reveal mistakes an AI assistant may not notice while working, such as ignoring a `CLAUDE.md` file or making claims without checking files.

check-on-claude — Command for Claude Code by john-broadway

Install

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.

Clone the repo

git clone --depth 1 https://github.com/john-broadway/maude-for-claude

Made for: Claude Code.

Or install maude, the plugin that ships this one along with the rest of its 1 skill, 15 commands, 1 agent, 8 hooks.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/commands/john-broadway/maude-for-claude/check-on-claude/github.svg)](https://agentmods.dev/commands/john-broadway/maude-for-claude/check-on-claude)

Your own site

<a href="https://agentmods.dev/commands/john-broadway/maude-for-claude/check-on-claude"><img src="https://agentmods.dev/badge/commands/john-broadway/maude-for-claude/check-on-claude/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.

Your own site · 80×15

<a href="https://agentmods.dev/commands/john-broadway/maude-for-claude/check-on-claude"><img src="https://agentmods.dev/badge/commands/john-broadway/maude-for-claude/check-on-claude.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 38 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 691 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

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	$0.00038	$0.00691
Opus 5	$0.00019	$0.00345
Sonnet 5	$0.00008	$0.00138
Haiku 4.5	$0.00004	$0.00069

Measured 11d ago against content hash f259ce84b993, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

check-on-claude 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 11d 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.

commands/check-on-claude.md · 59 lines

How it starts

The opening of the file, as written. The whole thing — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/maude:check-on-claude

You are Maude. Check on Claude — the partner you live with. He doesn't naturally watch his own behavior. You do.

What to do

PROJ="${CLAUDE_PROJECT_DIR:-$(pwd)}"
SLUG="$(printf %s "$PROJ" | sed 's/[^a-zA-Z0-9]/-/g')"
MEM="$HOME/.claude/projects/$SLUG/memory"
SELF="$PROJ/.maude/plugin"
USER_DIR="$HOME/.claude/maude"
REMEMBER="$PROJ/.remember"
TRACE="$SELF/trace/today-$(date -u +%Y-%m-%d).jsonl"   # UTC — matches the trace writer's filename clock

Repeated tool calls — read the trace. Bucket by tool + arguments. Flag anything called > 3 times with the same/similar input ("Claude grep'd for the same term four times today").
Unread context — has Claude read CLAUDE.md this session? Read the trace and look for Read calls with that path. If the project HAS a CLAUDE.md and Claude hasn't read it, that's a flag.
Stale house-map — when was the last /maude:found walk? If > 7 days and the workspace has visibly changed (ls + recent file mtimes), the map is stale.
Confabulation risk — look for assistant turns in the trace that made claims about file contents, paths, or APIs without a preceding Read/Grep tool call backing them. Hard to detect perfectly; surface as a soft signal: "Claude made N claims this session without a tool call to verify — worth a re-check?"
MEMORY.md drift — has the project's MEMORY.md changed since session start, and did Claude re-read it? If not, his loaded context may be stale.
Open loop check — does the trace show TaskCreate calls that were never marked completed? Pending todos he forgot?

Format

Claude check:
  Repeated calls: <N grep / N read / N bash, with the most-repeated highlighted>
  CLAUDE.md read this session: ✓ / ✗ (skipped — should read it)
  House-map: <fresh / stale (X days)>
  Open todos: <N pending>
  Confab risk: <low / N unbacked claims worth re-checking>

Suggestions for Claude:
  - <concrete>

Voice

Direct, but not punitive.
"Slow down."
"You haven't read CLAUDE.md. Read it."
"You've grep'd this same term four times — what are you actually looking for? Let me help."

Read the full file on GitHub · 59 lines

Changes

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.

11d ago First seen · 59 lines · 38 tokens per session scan A f259ce84b993

Subscribe to this mod's changes

check-on-claude is a command published in the GitHub repository john-broadway/maude-for-claude (2 stars, last pushed 8d ago), licensed Apache-2.0. It adds 38 tokens to every session and 691 once invoked, about $0.0002 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.

Building agents Claude Code ai-agent ai-pair-programming anthropic claude claude-code-plugin

recall-save

Generate / overwrite .recall/context.md with Recall's local offline summarizer.

raiyanyahya/recall · 15 tokens

security-audit

Comprehensive security audit of your project AND Claude Code configuration. Analyzes secrets exposure, injection surfaces, dependencies, hook security, and produces a scored security posture assessment.

FlorianBruniaux/claude-code-ultimate-guide · 0 tokens

version

Display current guide and Claude Code versions.

FlorianBruniaux/claude-code-ultimate-guide · 9 tokens

diff-docs

Compare official Anthropic docs baseline vs current snapshot (no network — instant).

FlorianBruniaux/claude-code-ultimate-guide · 15 tokens

notebook-query

Query the notebook knowledge base (SQLite) built by /agy:notebook — precise, grounded, cited. Ask in natural language ("sum the amounts by category", "which docs mention 'Acme Corp'", "build a project timeline") or pass raw SQL. Read-only. Use this when you need exact aggregates/lookups across a document corpus…

MarcosNahuel/antigravity-plugin-cc · 74 tokens

ccc-orchestrate

Sequential and tmux/worktree orchestration guidance for multi-agent workflows.

KevinZai/commander · 14 tokens