/cs-check-claims

Check an agent's summary, handoff or "tests pass" against what its recorded run actually did.

BOOST

Guide

alirezarezvani@alirezarezvani

GitHub

From plugin

claude-skills

30k169 skills127 agents169 commands2 MCP

$ npx -y skills add alirezarezvani/claude-skills --agent claude-code

Command definition

cs-check-claims.md

disable-model-invocation: true
description: Check an agent's summary, handoff or "tests pass" against what its recorded run actually did.
argument-hint: "[--session ID | --trace PATH] <claims or summary text>"

/cs-check-claims

Turn each factual sentence of the summary in `$ARGUMENTS` (default: your own last report in this session) into a typed claim and check it.

S=${CLAUDE_PLUGIN_ROOT}/skills/agent-run-forensics/scripts
python3 $S/claim_checker.py --trace <path> \
  --claim "passed:<test command>" --claim "edited:<file>" --claim "not-edited:<file>" --claim "ran:<command>"

Claim kinds: `ran:`, `not-run:`, `passed:`, `failed:`, `edited:`, `not-edited:`, `read:`.

Report

1. The verdict line. 2. Every CONTRADICTED claim first, with the step that contradicts it, as a correction. 3. STALE, PARTIAL, INFERRED and NOT_IN_TRACE claims, one line each. 4. Claims that held, briefly.

Read it on GitHub ↗

Stats

Python JavaScript HTML Bash GCP Git GitHub Playwright Markdown Anthropic Gemini MCP

29,589

Stars

4,104

Forks

Active

Maintenance

Python

Language

MIT

License

12h ago

Last commit

1y ago

Created

Repo: alirezarezvani/claude-skills

Ships withclaude-skills

411 production-ready skills for Claude Code and 12 other coding agents, covering engineering, product, marketing, compliance, finance and the C-suite. We test whether each skill loads when you ask in plain words, and publish the results.