agentleFS
brpazbrpaz/agent-skills/skills/bspec-technical-architecture/SKILL.md

bspec-technical-architecture

Skill1 starsChanged 6 months agoMITView on GitHub

Technical architecture skill for Bruno's personal spec workflow. Shapes docs/ARCHITECTURE.md, drives a feedback loop, and records durable decisions via ADRs when needed.

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 5 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill only when shaping or updating the canonical technical architecture.

    line 8
  2. Very specific:

    Create or refine a living architecture_document for work that is too risky, cross-cutting, or ambiguous to implement safely from the …

    line 12
  3. Very specific:

    start with status Draft

    line 62
  4. Distinctive:

    ask one high-leverage technical question at a time

    line 63
  5. Very specific:

    update the same architecture_document after each meaningful clarification

    line 65
  6. Distinctive:

    if implementation later discovers material architectural drift, update the same document instead of creating a parallel design doc

    line 68
  7. Very specific:

    use adr-writer when the architecture includes a durable technical decision

    line 118
  8. Distinctive:

    create or update ADRs for architecture boundaries, persistence strategy, integration strategy, auth or security boundaries, deployment …

    line 119
  9. Very specific:

    keep architecture_document cross-cutting, durable, and honest

    line 136
  10. Very specific:

    keep story-specific execution detail in implementation_plan, not here

    line 137
  11. Distinctive:

    use this document as living technical truth for implementation and review

    line 138

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Technical Architecture

Use this skill only when shaping or updating the canonical technical architecture.

Purpose

Create or refine a living architecture_document for work that is too risky, cross-cutting, or ambiguous to implement safely from the business PRD alone.

  • prd_document = business and product truth
  • architecture_document = canonical technical architecture truth
  • implementation_plan = story-specific execution detail for the active implementation
  • ADRs = durable architecture decisions extracted when they matter beyond one implementation pass

The architecture document should stay current as implementation evolves. If implementation materially changes architecture, the architecture document must be updated.

When To Use

  • tech stack selection or revision
  • API design and contracts
  • schema or persistence boundaries
  • auth, security, or permission boundaries
  • external integrations, queues, webhooks, cron jobs, or infrastructure choices
  • module or package structure
  • runtime, deployment, or operational constraints
  • cross-cutting abstractions or tradeoffs that several stories will rely on

Focus Areas

  • technical goals
  • non-goals
  • constraints
  • tech stack
  • system context and boundaries
  • modules and responsibilities
  • interfaces and contracts
  • data model and storage boundaries
  • integrations and external systems
  • deployment and runtime model
  • security and permission boundaries
  • observability and testing strategy
  • risks and tradeoffs
  • open technical questions
  • architecture change log

Avoid

  • rewriting product requirements
  • acceptance criteria duplication
  • story decomposition
  • vertical slice planning
  • per-story implementation sequencing
  • issue checklists or task tracking
  • feature-specific edge-case detail that belongs in implementation planning

Feedback Loop

  • start with status Draft
  • ask one high-leverage technical question at a time
  • challenge weak assumptions and shallow tradeoff analysis
  • update the same architecture_document after each meaningful clarification
  • move to In Review when the architecture is coherent enough for feedback
  • move to Approved only when the user explicitly approves it or clearly asks to issue work against it
  • if implementation later discovers material architectural drift, update the same document instead of creating a parallel design doc
  • while architecture work is pending, keep related issues in status/needs-tech-design; when architecture is stable, move them back to status/ready

Recommended Structure

# Technical Architecture

## Status
Draft

## Linked Artifacts
- PRD:
- ADRs:
- Relevant issues:

## Technical Goals

## Non-Goals

## Constraints

## Tech Stack

## System Context

## Module Structure

## API Design and Contracts

## Data Model and Storage Boundaries

## Integrations and External Systems

## Deployment and Runtime

## Security and Access Boundaries

## Observability and Testing Strategy

## Risks and Tradeoffs

## Open Questions

## Architecture Change Log
- YYYY-MM-DD: initial draft

ADR Rules

  • use adr-writer when the architecture includes a durable technical decision
  • create or update ADRs for architecture boundaries, persistence strategy, integration strategy, auth or security boundaries, deployment decisions, and consequential tradeoffs
  • link ADRs from the architecture document instead of copying them in full

Definition Of Ready

The architecture is ready for issue generation or implementation handoff when:

  • the tech stack and major dependencies are clear enough
  • the main module boundaries and responsibilities are explicit
  • the important APIs, contracts, and data boundaries are explicit
  • the major runtime, security, and integration constraints are known
  • the main risks and tradeoffs are explicit
  • durable decisions are captured in ADRs when needed
  • the document status is Approved or it has been explicitly updated to reflect accepted implementation changes

Rules

  • keep architecture_document cross-cutting, durable, and honest
  • keep story-specific execution detail in implementation_plan, not here
  • use this document as living technical truth for implementation and review

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

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.