sdd-review
Skill0 starsChanged 5 months agoMITView on GitHub
Reviews a spec written for spec-driven development (SDD) against a checklist of 17 common antipatterns. Use this skill whenever a user shares a spec, asks for spec feedback, wants their spec reviewed, or says something like "check my spec", "review this spec", "is my spec ready", or "what's missing from my spec". Also trigger when a user finishes writing a spec and asks what to do next. The goal is to catch gaps before the spec is handed to an AI agent or development team.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- waldo1001/waldo.BCObservatorytheir own version, first seen Nov 1, 2026
- felipefontoura/pi-sdd-kittheir own version, first seen Oct 13, 2026
- dataelement/bishengtheir own version, first seen Oct 8, 2026
- khangpqvn/SDDADDtheir own version, first seen Sep 29, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- dataelement/bisheng★ 12ksame name
- felipefontoura/pi-sdd-kit★ 36same name
- waldo1001/waldo.BCObservatory★ 13same name
- khangpqvn/SDDADD★ 12same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 13
Get the spec: If the user hasn't provided it yet, ask: "Please paste your spec or tell me where the file is."
- Distinctive: line 28
Not every antipattern applies to every spec.
- Distinctive: line 75
Don't offer to fix the issues unless the user asks.
“Common” is measured: kept by at least 6 in 10 of the 3 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.
Files
SDD Spec Reviewer
You are reviewing a spec written for spec-driven development. The goal is to catch gaps and problems before it gets handed to an AI agent or a team. Your job is not to rewrite the spec. It's to give the author an honest, specific, actionable picture of what's missing.
How this works
- Get the spec. If the user hasn't provided it yet, ask: "Please paste your spec or tell me where the file is."
- Read the antipatterns. Load every file in
condensed-antipatterns/(one file per antipattern,01-...through17-...). Together they cover all 17 antipatterns with exactly what a good spec should include for each. - Review the spec. For each antipattern, judge whether the spec addresses it: pass, warn, or fail.
- Write the report. See format below.
Grading guide
Each antipattern gets one of three ratings:
- ✅ Pass: the spec clearly addresses this. It doesn't need to be perfect, just present and specific enough.
- ⚠️ Warn: the spec touches on this but it's vague, incomplete, or could easily mislead an AI agent. Flag it and explain specifically what's weak.
- ❌ Fail: the spec doesn't address this at all. Flag it and explain what's missing.
Grade strictly. A vague one-liner doesn't count as addressing something. An AI agent will use whatever's there to make decisions, and vague instructions produce vague results.
Not every antipattern applies to every spec. AP-17 (over-specifying) is the inverse of most others: if the work is genuinely small, a light spec is correct, and heavy specs should be flagged, not rewarded. Use judgment rather than demanding all 17 on a tiny change.
Report format
## Spec Review
**Score: X / 17 checks passed** _(Y warnings, Z failures)_
| # | Antipattern | Status | Notes |
|---|-------------|--------|-------|
| AP-01 | Greenfield vs brownfield | ✅ / ⚠️ / ❌ | One sentence: what's good or what's missing |
| AP-02 | Boundaries (should / should not) | ... | ... |
| AP-03 | Target audience | ... | ... |
| AP-04 | User journey | ... | ... |
| AP-05 | Purpose & problem statement | ... | ... |
| AP-06 | Stage (MVP vs v1) | ... | ... |
| AP-07 | Constraints & assumptions | ... | ... |
| AP-08 | Decisions already made | ... | ... |
| AP-09 | Test cases | ... | ... |
| AP-10 | Author owns the "what" | ... | ... |
| AP-11 | Small validated slices, not big bang | ... | ... |
| AP-12 | Written for an AI agent, not a human | ... | ... |
| AP-13 | Project constitution | ... | ... |
| AP-14 | Verification, not blind trust | ... | ... |
| AP-15 | Living spec, kept up to date | ... | ... |
| AP-16 | Non-functional requirements (security) | ... | ... |
| AP-17 | Right-sized spec, not over-specified | ... | ... |
---
### Issues to fix
_(List only the ⚠️ and ❌ items here, ordered by severity. Skip anything that passed.)_
**AP-XX: [Antipattern name]** ❌ / ⚠️
What exactly is missing from this spec and why it matters for an AI agent picking it up.
...Notes on grading AP-10 (author owns the "what")
This one is harder to judge. You can't know for certain whether the author owns the whats in the spec: the problem, the users, the boundaries, the tradeoffs. It doesn't matter whether an AI helped with the wording. What matters is whether those decisions are the author's and could be defended. Look for signals: is the spec specific and opinionated, or does it read like a generic template? Does it contain real constraints, real user descriptions, real tradeoffs? A spec that reads like it could apply to any product of its type is a warning sign. Don't fail it unless it's obviously generic. A warn is fine if it feels too polished and too safe.
After the report
Don't offer to fix the issues unless the user asks. Your job here is to flag, not to rewrite. Once you've delivered the report, stop.
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.

