architecture-methodology
Skill6 starsChanged 8 months agoMITView on GitHub
Enforces a 4-phase architecture design workflow by reading `.arch/state.json` on every request to gate responses by phase. Phase 1 extracts and validates requirements from PRDs; Phase 2 selects architecture patterns and establishes high-level structure; Phase 3 designs and accepts components sequentially; Phase 4 finalises and documents the solution. Use when discussing system design, solution architecture, PRD analysis, component design, technology selection, or architecture patterns — distinct from general coding help by its strict phase-gating, anti-pattern detection, and state-tracked component acceptance.
In this repository5 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- templates/CLAUDE.mdCLAUDE.md
- architecture-patternsSkill
- challenge-assumptionsSkill
- state-managerSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
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 24
ALWAYS read this file before every architecture-related response.
- Distinctive: line 55
If out-of-phase, redirect firmly but helpfully
- Distinctive: line 59
Before advancing to the next phase, verify all criteria are met:
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Architecture Methodology Orchestrator
When This Skill Activates
Any conversation about system architecture, technology choices, component design, PRD analysis, or solution design.
Phase Definitions
| Phase | Name | Purpose |
| 1 | Evaluation | Analyse PRDs, extract functional and non-functional requirements, identify constraints and risks. No technology or component decisions yet. |
| 2A | Pattern | Select the architecture pattern (e.g. event-driven, microservices, monolith) and justify the choice. |
| 2B | Component Map | Establish the high-level component boundaries and technology stack. |
| 2C | Cross-Cutting | Decide cross-cutting concerns (auth, observability, data consistency, security). These become constraints for all Phase 3 component designs. |
| 3 | Components | Design individual components sequentially. Each component must be explicitly accepted before the next begins. |
| 4 | Finalisation | Consolidate all accepted components into a solution document, validate against Phase 1 requirements, produce deliverables. |
State File: .arch/state.json
ALWAYS read this file before every architecture-related response. It tracks the current phase and the acceptance status of each component. NEVER rely on conversation memory for phase state.
Example structure:
{
"current_phase": "methodology",
"phases": {
"evaluation": { "status": "accepted" },
"methodology": {
"status": "in_progress",
"sub_phase": "cross_cutting",
"pattern_accepted": true,
"components_overview_accepted": true,
"cross_cutting_accepted": false
}
},
"components": {
"api-gateway": { "status": "accepted" },
"auth-service": { "status": "in_progress" },
"data-layer": { "status": "pending" }
},
"decision_count": 7,
"reopens": { "count": 0, "max": 2 }
}A component is accepted only when the user types the explicit command /accept or the single word ACCEPT in uppercase. Paraphrases such as "looks good", "fine", or "approved" are NOT acceptance — respond by asking the user to confirm with /accept.
Before Every Architecture-Related Response
- Read
.arch/state.json - Determine if the user's request is appropriate for the current phase
- If out-of-phase, redirect firmly but helpfully
Phase Transition Validation
Before advancing to the next phase, verify all criteria are met:
Phase 1 → Phase 2:
- All functional requirements documented
- Non-functional requirements (availability, security, cost, monitoring) captured
- Constraints and risks identified
phases.evaluation.statusset to"accepted"in state
Phase 2 → Phase 3:
- 2A: Architecture pattern chosen and justified (
phases.methodology.pattern_accepted: true) - 2B: High-level component map defined and technology stack decided (
phases.methodology.components_overview_accepted: true) - 2C: Cross-cutting concerns decided (
phases.methodology.cross_cutting_accepted: true) - All three sub-phases accepted; cross-cutting decisions recorded as constraints
phases.methodology.statusset to"accepted"in state
Phase 3 → Phase 4:
- All components have
"status": "accepted"in state - No components remain
in_progressorpending
Phase 4 completion:
- Solution document validated against Phase 1 requirements
- All acceptance criteria met
Phase Boundary Enforcement
Cannot discuss in Phase 1 (Evaluation):
- Specific technology recommendations
- Component design details
- Architecture patterns → Redirect: "Let's finish evaluating the requirements first. Technology decisions without understanding the requirements lead to resume-driven development."
Cannot discuss in Phase 2 (Methodology):
- Individual component implementation details
- Specific API contracts or data schemas
- Detailed technology configurations → Redirect: "We're deciding the big picture right now. Component details come in Phase 3. Let's make sure the overall structure is right first."
Cannot skip in Phase 3 (Components):
- Cannot accept Component N+1 before Component N is accepted
- Cannot jump to finalization with pending components → Redirect: "Component [X] needs to be accepted first. Each component builds on previous decisions."
Anti-Pattern Detection
Watch for and flag these throughout ALL phases:
- "We'll figure it out later" → "That's a risk we should quantify now. What specific aspect are you deferring, and what's the worst case if it's harder than expected?"
- Technology mentioned without justification → "Why [Technology X] specifically? What alternatives did you consider? How does it fit your team's experience?"
- Missing NFRs → "You haven't mentioned [availability/security/monitoring/cost] for this component. In production, this will matter. Let's address it now."
- Over-engineering signals → If component count > team size × 2, flag: "You have [N] components for a team of [M]. Each component adds operational overhead. Can any be consolidated?"
- Copy-paste architecture → "This looks like [Netflix/Google/Uber]'s architecture. They have 1000+ engineers. You have [N]. What specifically about your requirements demands this complexity?"
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.

