Reconcile the REAL project against the learn-as-you-go guide you have been following — does what you built actually contain what the steps you executed told you to write? Reads the milestone checkpoints (the complete-file sections of each NN_verify.md), diffs them against your files while ignoring cosmetic differences, cross-checks foundation/progress.md against version-control history to catch steps ticked but never performed, re-runs only the gate checks that need no human, and classifies e...
Scanned 9/4/2026
Install to Claude Code
npx -y skills add tomthebearded/guide-forge --skill check-my-work --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Check My Work?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/tomthebearded-check-my-work)More formats (shields.io, HTML) on the badges page.
---
name: check-my-work
description: >
Reconcile the REAL project against the learn-as-you-go guide you have been following — does what you built
actually contain what the steps you executed told you to write? Reads the milestone checkpoints (the
complete-file sections of each NN_verify.md), diffs them against your files while ignoring cosmetic
differences, cross-checks foundation/progress.md against version-control history to catch steps ticked but
never performed, re-runs only the gate checks that need no human, and classifies every difference as
cosmetic / deliberate / defect / never executed. Read-only: it changes neither the project nor the guide.
Use when you doubt you followed the guide correctly, before a milestone gate, or when you ask to mark a run
of steps or a whole milestone done and want the claim checked before /mark-progress records it. Invoke with
the scope, e.g. "/check-my-work M6" or "/check-my-work everything I've executed".
argument-hint: "<step|milestone|range> [file ...]"
---
# Skill: check-my-work
The complete, canonical contract for this skill is the paste-prompt it wraps, inlined here — **follow it in
full:**
!`cat "${CLAUDE_SKILL_DIR}/prompt.md"`
(If the block above came back empty or errored, read `prompt.md` in this skill folder — it sits beside this `SKILL.md` — and follow that.)
As you follow it, adapt for the fact that you're a Claude Code skill, not a pasted prompt:
- **You're already invoked** — skip its "How to use / paste this" preamble. The scope to check is
`$ARGUMENTS` below (plus any attached files). Read the guide's `guide/` files and **the project's real
files** yourself rather than asking for them to be pasted.
- **You have tools here, so use the two the prompt leans on:** read version control directly (which commits
touched which files, and when) to answer *was this step ever performed*; and actually **run** the gate
checks that need no human, then diff their output against the expected output the gate prints.
- **Read-only, on both sides.** Do not edit the project, and do not edit any guide file — including
`progress.md` and `status.md`. If the check ends in something that must be written, name the sibling
**skill** that owns it: `/mark-progress` (the ledger), `/report-issue` (a guide defect), `/amend-guide`
(a deliberate deviation that should become the plan), `/log-feedback` (friction worth recording).
- When the reader asked to **mark** a run of steps or a whole milestone, check first and hand the confirmed
result to `/mark-progress` — a single step just finished doesn't need this skill at all.
Scope to check / args:
```
$ARGUMENTS
```
**Gate:** report only what you actually verified. A check you couldn't run is **unrun**, never passed; a
difference you can't tell apart from a rename is **cosmetic**, not a finding; and a milestone is never called
verified here, because you did not watch its gate — you re-ran the half of it a machine can.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!