spec-proofreader
Skill1 starsChanged 39 days agoMITView on GitHub
Reviews spec documents for text quality issues without knowledge of the feature description. Fixes errors directly. Invoked by spec-writer, not directly.
In this repository38 other files it gives its agents.
- agent-conductSkill
- bugfixSkill
- code-smellsSkill
- deliverSkill
- final-responseSkill
- frontend-designSkill
- go-conventionsSkill
- go-implementorSkill
- go-reviewerSkill
- go-test-strengthSkill
- implementation-principlesSkill
- nextflow-conventionsSkill
- nextflow-implementorSkill
- nextflow-reviewerSkill
- nextjs-fastapi-conventionsSkill
- nextjs-fastapi-implementorSkill
- nextjs-fastapi-reviewerSkill
- orchestratorSkill
- phase-creatorSkill
- phase-reviewerSkill
- prose-principlesSkill
- pr-resolverSkill
- pr-reviewerSkill
- python-conventionsSkill
- python-implementorSkill
- python-reviewerSkill
- spec-authorSkill
- spec-reviewerSkill
- spec-writerSkill
- subagentsSkill
- swift-conventionsSkill
- swift-implementorSkill
- swift-reviewerSkill
- testing-principlesSkill
- tightenSkill
- unslopSkill
- verificationSkill
- writing-for-agentsSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
13 rules. About 0 are common, 12 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈12
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Read and follow agent-conduct and unslop before starting.
- Distinctive: line 15
Spec path: only. Do NOT receive the feature description.
- Distinctive: line 28
AI tells: every rule in prose-principles, applied to the spec's prose.
- Distinctive: line 34
No gaps or duplicates.
- Distinctive: line 35
Every story ID in implementation order exists as an actual story.
- Distinctive: line 36
Every story appears in the implementation order.
- Distinctive: line 37
All acceptance tests are inside user story blocks.
- Very specific: line 38
All stories are under matching ## Section <Letter>: headings.
- Distinctive: line 52
PASS: - no errors found.
- Distinctive: line 57
Do NOT check feature coverage (that is spec-reviewer's job).
- Distinctive: line 58
Do NOT evaluate technical design.
- Distinctive: line 59
Do NOT add new content.
- Distinctive: line 60
ONLY fix text errors, formatting, numbering, and consistency.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Spec Proofreader Skill
Read and follow agent-conduct and unslop before starting.
You have NO knowledge of the original feature description. Review the spec purely on its own merits as a written document.
Input
- Spec path only. Do NOT receive the feature description.
Procedure
1. Read the entire spec.
2. Check for text errors
- Repetition: the same thing in different words, or redundant acceptance tests.
- Contradictions: conflicting statements (e.g. type defined differently in two places).
- Undefined terms: types, functions, packages used without definition.
- AI tells: every rule in prose-principles, applied to the spec's prose.
3. Check structure and numbering
- Sequential section letters (A, B, C...) and story numbers (A1, A2, B1...). No gaps or duplicates.
- Every story ID in implementation order exists as an actual story.
- Every story appears in the implementation order.
- All acceptance tests are inside user story blocks.
- All stories are under matching
## Section <Letter>:headings.
4. Check formatting
Apply unslop § Mechanics, which covers wrapping, placeholder text, whitespace, code fences, headings, and cross-references. The spec adds one rule: ASCII only outside code blocks, so no section marks either.
5. Fix errors directly. Keep fixes minimal. Check codebase only to resolve
ambiguities.
6. Verdict
- PASS - no errors found.
- FIXED - list every error and how it was fixed.
Rules
- Do NOT check feature coverage (that is spec-reviewer's job).
- Do NOT evaluate technical design.
- Do NOT add new content.
- ONLY fix text errors, formatting, numbering, and consistency.
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
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.

