agentleFS
adezdevadezdev/blueprint/skills/inspect/SKILL.md

inspect

Skill0 starsChanged 3 months agoMITView on GitHub

Validate a Blueprint feature's traceability chain — orphan requirements, uncovered acceptance criteria, dangling C/R ids, components nothing builds, unanswered open questions. Use when the user says "/blueprint:inspect", "check the spec", "validate the blueprint", "does everything trace", or after hand-editing any of requirements.md / architecture.md / tasks.md.

13 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read ${CLAUDE_PLUGIN_ROOT}/FORMAT.md first (or FORMAT.md at the repo root if you are working inside the blueprint repo itself).

    line 8
  2. Distinctive:

    Read-only. This skill reports.

    line 11
  3. Very specific:

    Run these against .blueprint/<slug>/ and work from the output — do not eyeball the documents:

    line 16
  4. Distinctive:

    If a document is missing, report which step has not run yet and stop — do not report 40 failures because architecture.md does not exist.

    line 48
  5. Very specific:

    Read declarations, not token occurrences.

    line 51
  6. Distinctive:

    When everything passes, say so in one line with the counts.

    line 78

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Blueprint: inspect

Read ${CLAUDE_PLUGIN_ROOT}/FORMAT.md first (or FORMAT.md at the repo root if you are working inside the blueprint repo itself).

Read-only. This skill reports; it does not edit. Offer fixes, apply none without being asked.

Gather

Run these against .blueprint/<slug>/ and work from the output — do not eyeball the documents:

rg -o '^### (R\d+)'            requirements.md   # declared requirements
rg -o '\*\*(R\d+\.AC\d+)\*\*'  requirements.md   # declared criteria
rg -o '^### (C\d+)'            architecture.md   # declared components
rg -o 'covers: *(.+)$'         architecture.md   # coverage claims
rg -o '\*\*(T\d+)\*\*'         tasks.md          # declared tasks
rg -o '→ *(C\d+|chore) *\| *(.+)$' tasks.md      # task citations
rg -o '\*\*(Q\d+)\*\*'         requirements.md architecture.md
rg -n 'status: answered'       requirements.md architecture.md
rg -n 'status: dropped'        requirements.md architecture.md tasks.md

Checks

#CheckSeverity
1Every R<n> has ≥1 ACfail
2Every R<n> is covered — named directly in a covers: list, or every one of its criteria isfail
3Every R<n>.AC<m> is cited by ≥1 taskfail
4Every C<n> is cited by ≥1 taskfail
5Every id cited anywhere actually exists (no dangling refs)fail
6No id declared twicefail
7Nothing references an id marked status: droppedfail
8Every task has a done-when linefail
9Every task has a files: linewarn
10No unanswered Q<n> — rg '^\s*- \*\*Q\d+\*\* status: answered' matches every declared Qwarn
11No task cites more than one C<n>warn
12Phases exist and each has ≥1 taskwarn

If a document is missing, report which step has not run yet and stop — do not report 40 failures because architecture.md does not exist.

Read declarations, not token occurrences. These documents describe Blueprint's own grammar, so files:, status: dropped and covers: all turn up inside ordinary prose. A marker counts only where the format puts it: at the start of its line, in the document that owns it. A phantom id lifted out of a sentence is worse than a missed one, because it sends the reader hunting for something that was never declared.

Output

A short table, then the specifics. Nothing else.

Blueprint: token-refresh

  requirements   7 R, 19 AC
  architecture   5 C
  tasks          12 T, 3 phases

  FAIL  R4 covered by no component
  FAIL  T9 cites C6, which does not exist
  FAIL  R2.AC3 cited by no task
  WARN  Q2 unanswered
  WARN  T5 has no files: line

  3 failures, 2 warnings.

When everything passes, say so in one line with the counts. Do not pad.

For each failure, name the cheapest fix in a few words ("R4: add a component or drop the requirement"). Then ask whether to apply them. Do not start editing.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.