architecture-review
Skill1 starsChanged 5 months agoMITView on GitHub
Run a structured architecture decision process using subagent personas that debate the decision, then produce a signed ADR (markdown + optional HTML) and update a decision registry. Triggers on "should I use X or Y", "architecture decision", "ADR", "help me decide between", "which approach", "debate this", "tech decision". Use whenever the user is making a technical or architectural choice and needs structured thinking, not just a list of tradeoffs.
85 other projects use a version of this.
Adopted by
Most recently taken up by
- lx-wnk/skillstheir own version, first seen Nov 16, 2026
- seancabalse/agentic-playbooktheir own version, first seen Nov 16, 2026
- erikperttu/steering-coding-agentstheir own version, first seen Nov 16, 2026
- Sequela02/canastatheir own version, first seen Nov 15, 2026
- OmarAlex24/audit-skillstheir own version, first seen Nov 15, 2026
- chuanglibs/Claude-Code-Game-Studios-Codextheir own version, first seen Nov 15, 2026
Used with
Projects with it also use create-stories, story-done and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
lilangMaxSkillcreate-stories
Break a single epic into implementable story files. Reads the epic, its GDD, governing ADRs, and control manifest. Each story embeds its GDD requirement TR-ID, ADR guidance, acceptance criteria, story type, and test evidence path. Run after /create-epics for each epic.
Also in 11 projects that use this
lilangMaxSkillstory-done
End-of-story completion review. Reads the story file, verifies each acceptance criterion against the implementation, checks for GDD/ADR deviations, prompts code review, updates story status to Complete, and surfaces the next ready story from the sprint.
Also in 11 projects that use this
- +10 more from lilangMax
Other versions85 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 85
- Donchitos/Claude-Code-Game-Studios★ 39ksame name
- langwatch/langwatch★ 5ksame name
- ghuntley/loom★ 1.7ksame name
- microsoft/hve-core★ 1.6ksame name
- DataDog/dd-trace-js★ 842same name
- decebals/claude-code-java★ 763same name
- borgbase/vykar★ 726same name
- appdevforall/CodeOnTheGo★ 621same name
- owainlewis/blueprint★ 432same name
- getsentry/warden★ 424same name
75 more not listed
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.
In this repository51 other files it gives its agents.
- architect-enterprise-aiSkill
- author-agent-skillsSkill
- build-agent-harnessesSkill
- build-aiewf-2024-ragSkill
- build-generative-mediaSkill
- build-multimodal-productsSkill
- build-rag-search-stacksSkill
- build-robotics-aiSkill
- build-with-deepmindSkill
- build-with-microsoft-aiSkill
- build-with-openaiSkill
- deploy-public-sector-aiSkill
- design-agent-skillsSkill
- evaluate-with-braintrustSkill
- lead-ai-engineering-teamsSkill
- lead-ai-transformationSkill
- learn-summit-2023-remoteSkill
- learn-summit-2023-talksSkill
- navigate-aie-europe-2026Skill
- navigate-aiewf-2024-keynotesSkill
- observe-ai-productionSkill
- operate-agent-memorySkill
- operate-legal-ai-agentsSkill
- operate-openclaw-agentsSkill
- optimize-llm-inferenceSkill
- plan-sovereign-aiSkill
- price-ai-productsSkill
- run-aiewf-2024-evalsSkill
- run-llm-evalsSkill
- run-summit-2023-workshopsSkill
- secure-ai-agentsSkill
- ship-ai-as-pmSkill
- ship-aiewf-2024-agentsSkill
- train-llms-locallySkill
- chat-failure-auditSkill
- create-a-skillSkill
- cross-model-reviewSkill
- cto-weekly-reviewSkill
- friction-auditSkill
- ig-carouselSkill
- pdf-reportSkill
- raise-prSkill
- reel-editSkill
- research-reportSkill
- ship-checkSkill
- spec-grounded-designSkill
- state-lifetime-decisionSkill
- ste100Skill
- scroll-engineeringSkill
- ui-uxSkill
- verify-like-a-userSkill
The file
We split this skill into the rules it gives an agent
11 rules. 0 are common, 11 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 77 edited versions we read keep these.
11
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 26
Ask the user (in one message, all at once):
- Distinctive: line 34
Do not proceed until you have at least 1, 2, and 3.
- Distinctive: line 45
Create the output directory if it does not exist.
- Distinctive: line 67
Read personas.md. Run subagents in parallel when supported.
- Distinctive: line 78
Keep subagent contexts isolated — no persona sees another's response until synthesis.
- Distinctive: line 91
After all persona subagents return, synthesize using the Synthesis prompt in personas.md.
- Distinctive: line 94
Preferred: Run a synthesis subagent with all persona responses as context.
- Distinctive: line 102
Always save a markdown ADR.
- Distinctive: line 201
Subagents unavailable: run personas sequentially in isolated turns, or simulate each persona before synthesizing (note lower quality to …
- Distinctive: line 202
Custom output dir missing: create it; warn if path is outside the project
- Distinctive: line 203
Re-run same decision: write a new ADR number; mark the old ADR Status as Superseded with a link to the new one
“Common” is measured: kept by at least 6 in 10 of the 77 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
Architecture Decision Record — Subagent Debate Skill
Structured persona debate → synthesis → markdown ADR (+ optional HTML report) → registry index.
Works with any coding agent that can run subagents. No external API calls required.
Supporting files (read when needed):
- personas.md — persona and synthesis prompts
- adr-template.md — markdown ADR output format
Step 0 — Elicit the Decision
Ask the user (in one message, all at once):
- What is the decision? (e.g. "FTS5 vs pgvector for legal document search")
- What are the hard constraints? (latency, cost, team size, existing infra, timeline)
- What is the context? (stage of product, scale, who will maintain this)
- Do you have a leaning already? (optional — being honest about priors helps)
- What does "good" look like? (what would make this decision clearly correct in 6 months)
Do not proceed until you have at least 1, 2, and 3.
Also resolve output settings (infer from request if stated, else use defaults). See ../OUTPUT.md for the shared convention across all skills.
| Setting | Resolution |
| Output directory | {SKILL_OUTPUT_DIR}/architecture-review/ — default root ./skill-outputs/ |
| Include SRE persona | Yes if user asks, or decision touches deployment/monitoring/on-call/scaling/data durability/managed-vs-self-hosted |
| HTML report | Yes by default; skip if user says "ADR only", "markdown only", or "no report" |
Create the output directory if it does not exist.
Step 1 — Brief All Subagents
Construct a decision brief from the user's answers:
DECISION: [what is being decided]
CONSTRAINTS: [hard constraints]
CONTEXT: [product stage, scale, team, infra]
OPTIONS: [Option A] vs [Option B] (vs [Option C] if applicable)
SUCCESS CRITERIA: [what good looks like]
USER LEANING: [if stated, else "none stated"]
INCLUDE_SRE: [yes/no]Step 2 — Run the Debate
Read personas.md. Run subagents in parallel when supported.
Always run (3 subagents): Pragmatist, Architect, Contrarian
Optionally run (4th subagent): SRE — when INCLUDE_SRE: yes
Pass each subagent:
- The decision brief from Step 1
- The matching persona prompt from
personas.md - Instruction: respond using the format in the persona prompt
Keep subagent contexts isolated — no persona sees another's response until synthesis.
| Persona | Icon | HTML color |
| Pragmatist | ⚙️ | #111111 |
| Architect | 🏛 | #4a5568 |
| Contrarian | ⚡ | #744210 |
| SRE | 🔧 | #2c5282 |
Step 3 — Synthesis Pass
After all persona subagents return, synthesize using the Synthesis prompt in personas.md.
Preferred: Run a synthesis subagent with all persona responses as context.
Fallback: Synthesize yourself as the orchestrating agent using the same format.
Step 4 — Write Markdown ADR
Always save a markdown ADR. Fill in adr-template.md from synthesis output and persona responses.
Path: {output_dir}/ADR-NNNN-[decision-slug].md
ADR numbering: Scan {output_dir}/ for existing ADR-*.md files. Use the next
sequential number (ADR-0001, ADR-0002, …). Start at ADR-0001 if none exist.
The date lives inside the document, not the filename.
Step 5 — Generate HTML Report (optional)
Skip if user chose ADR-only mode.
Path: {output_dir}/ADR-NNNN-[decision-slug].html (same NNNN and slug as the .md)
Link to the markdown ADR from the HTML header.
Design spec
Page bg: #f8f7f5
Card bg: #ffffff
Border: 1px solid #e8e5de
Primary text: #111111
Muted: #6b6b6b
Hint: #a8a39a
Pragmatist: #111111
Architect: #4a5568
Contrarian: #744210
SRE: #2c5282
Synthesis: #1a3a2a
Font: Georgia/serif for debate sections + system-ui for chrome
NO gradients. NO purple. Light and premium.Report sections
HEADER — "Architecture Decision Record" · date · link to .md · "Export PDF" (print)
DECISION TITLE — large, prominent
CONTEXT STRIP — 3 pills: constraints, product stage, success criteria
THE DEBATE — columns per persona (3 or 4). Each column: icon, POSITION, ARGUMENT, risks/ops burden, CONFIDENCE badge, CAVEAT in italics.
TRADEOFF MATRIX — rows = evaluation dimensions; columns = options; cells = ✓ / ~ / ✗ + note
DECISION DIAGRAM — SVG/CSS comparison cards with CHOSEN highlight
SYNTHESIS — background #f0ede8: VERDICT, REASONING, agreed/decisive/dissent blocks, CONDITIONS
SIGNED ADR — monospace card with all ADR fields
FOOTER — Architecture Decision Record · generated [timestamp]
Step 6 — Update Decision Registry
Maintain {output_dir}/index.md as a chronological registry of all ADRs.
If index does not exist, create:
# Architecture Decision Records
| ADR | Date | Status | Decision | Verdict |
|-----|------|--------|----------|---------|After each new ADR, append a row (do not duplicate if re-running the same file):
| [ADR-NNNN: Title](ADR-NNNN-decision-slug.md) | YYYY-MM-DD | Accepted | [options compared] | [one-line verdict] |Sort rows by date descending (newest first). If updating an existing ADR in place, update its row rather than appending a duplicate.
Step 7 — Output to User
- List saved file paths (
.md,.htmlif generated,index.mdupdated) - State the verdict in one sentence
- Name the decisive factor
- Note the review trigger
- Mention whether SRE persona participated
Edge Cases
- More than 2 options: give all options to every persona
- Strong user leaning: tell the Contrarian explicitly; ask them to challenge it
- Ambiguous decision: output a decision rule ("use A if X, use B if Y"), not a fake pick
- Re-run request: "argue the other side" — re-run debate with swapped priors; write a new ADR or update status to Superseded on the old one
- Subagents unavailable: run personas sequentially in isolated turns, or simulate each persona before synthesizing (note lower quality to user)
- Custom output dir missing: create it; warn if path is outside the project
- Re-run same decision: write a new ADR number; mark the old ADR Status as Superseded with a link to the new one
Invocation Examples
"should I use FTS5 or pgvector for legal search"
"ADR: Electron vs web app for our desktop client"
"help me decide between Redis and BullMQ for job queuing"
"debate this: do we build our own auth or use Clerk"
"architecture decision — monorepo vs separate repos"
"include SRE — managed Postgres vs self-hosted"
"ADR only"
"save to skill-outputs/architecture-review"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.

