Verify spec
This skill should be used when the user asks to "verify spec auto-format-config", "check acceptance criteria for pr-review-rereview", "is pr-review-ado-fetcher-reliability ready to merge", "run verify-spec on unic-archon-dlc", or invokes /verify-spec with an issue slug or partial name. Locates the matching PRD file in docs/issues/ and checks every acceptance criterion against the current codebase.From its SKILL.md
npx -y skills add unic/unic-agents-plugins --skill verify-specAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
4.1 KB, as published. Nobody here has run it
Purpose
Verify that all acceptance criteria in a spec file are satisfied by the current state of the codebase. Report each criterion as ✓ PASS or ✗ FAIL with a one-line evidence note, then summarise whether the spec is ready to be marked done.
Argument
$ARGUMENTS is one of:
- A full issue slug:
auto-format-config - A partial slug:
auto-format(substring match) - A done sub-task filename:
01-end-to-end-notice-pipeline.md
Locating the spec file
Scope limit: this skill reads spec files only. Since upstream v1.1, /to-spec publishes the spec as a GitHub issue and writes no file, so a Feature specced that way has nothing here to verify. Say so plainly — "this Feature's spec is a GitHub issue, which this skill cannot read yet" — rather than reporting no match, which reads as a missing spec.
Spec files live in docs/issues/ as <slug>/PRD.md. Completed vertical-slice sub-tasks live under docs/issues/<slug>/done/NN-title.md.
Search in order:
docs/issues/at the repo root — match<slug>/PRD.mdapps/claude-code/<any-plugin>/docs/issues/— same structure
Matching rules:
- Full slug (
auto-format-config): look fordocs/issues/auto-format-config/PRD.md. - Partial slug (
auto-format): substring match against slug directory names; list all matches and ask which to use if more than one. - Done sub-task filename (
01-end-to-end-notice-pipeline.mdor bare01): searchdone/sub-directories under matching slugs.
If no file matches, report that clearly and stop.
Spec file format
Spec files follow this structure — consult references/spec-format.md for full field definitions:
# NN. Title
**Status: open**
**Priority:** … **Effort:** … **Version impact:** … **Depends on:** … **Touches:** …
## Context
## Current behaviour
## Target behaviour
## Affected files
## Implementation steps
## Verification
## Acceptance criteria
## Out of scope
If the spec's Status is done, note it prominently at the top of the output, then proceed with full verification anyway — the codebase may have regressed since the spec was closed, and the purpose of this skill is to verify facts, not trust metadata.
The section to verify is ## Acceptance criteria. Read every line-item under it before starting verification.
Verifying each criterion
Run all criteria before reporting — do not stop on the first failure.
File existence: Use ls or Read to confirm the path exists.
File content: Use Read or grep to confirm text, fields, or values are present.
CLI check: Run the stated command (e.g. pnpm test, pnpm ci:check) and treat a non-zero exit as FAIL.
Negative check ("X should NOT exist"): Confirm absence with ls or grep.
Structural check (JSON/YAML field present): Read the file and inspect the field.
Behavioral check (a function/hook does X): Read the implementation and reason about it; mark PASS only when clearly satisfied.
When a criterion is ambiguous, make a best-effort determination and note the ambiguity in the evidence line.
Output format
For each criterion, output exactly:
✓ PASS <criterion text> — <one-line evidence>
✗ FAIL <criterion text> — <one-line gap description>
Then a summary block:
## Summary
Passed: N / Total: M
[Ready / NOT ready] to mark done.
Remaining gaps:
- <FAIL item 1 with specific gap>
- <FAIL item 2 with specific gap>
Only include "Remaining gaps" if there are failures. Do not mark the spec done if any criterion fails.
Additional resources
references/spec-format.md— annotated spec template with field definitions