| name | darwin-architect-bootstrap |
| description | Workspace scanning and structured analysis brief generation for unfamiliar repositories |
| modes | ["plan","review","analyze"] |
| roles | ["architect"] |
Workspace Bootstrap
When dispatched to analyze an unfamiliar repository or assess a change's impact, scan before planning. The goal is to produce a structured analysis brief that gives FRIDAY and downstream agents enough context to act.
Phase 1: Discovery
Read the repository's orientation signals first. These files establish intent, conventions, and deployment context:
- README / AGENTS.md / CONTRIBUTING.md -- project purpose, onboarding, AI-specific conventions
- Package manifests --
package.json, requirements.txt, go.mod, Cargo.toml, pyproject.toml
- CI configuration --
.github/workflows/, .tekton/, .gitlab-ci.yml, Jenkinsfile
- Deployment manifests --
helm/, kustomize/, Dockerfile, Containerfile, deploy/
- Directory structure -- top-level organization reveals monorepo vs single-service, frontend/backend split
Phase 2: Stack Identification
Classify the technology stack along these dimensions:
- Language(s) and runtime -- version constraints, package manager
- Framework(s) -- web framework, test framework, ORM
- Build system -- bundler, compiler, task runner
- Deployment target -- Kubernetes, serverless, VM, container registry
Report the stack as a concise summary table, not a prose paragraph.
Phase 3: Boundary Mapping
Identify the system's architectural surfaces:
- Ports and adapters -- where does external I/O enter/exit? (HTTP handlers, gRPC services, message consumers, CLI entrypoints)
- Service boundaries -- if multi-service, how do they communicate? (REST, gRPC, events, shared database)
- API surfaces -- public endpoints, internal endpoints, webhook receivers
- Data stores -- databases, caches, queues, object storage
Understanding boundaries prevents plans from accidentally crossing service ownership lines.
Phase 4: Convention Detection
Observe the patterns the repository already follows. Plans that respect existing conventions are adopted faster:
- Naming -- file naming, variable casing, module organization
- Error handling -- centralized vs local, error types, logging patterns
- Testing -- test location, framework, coverage expectations, fixture patterns
- File organization -- flat vs nested, feature-based vs layer-based
Report conventions as observed facts, not recommendations.
Phase 5: Brief Generation
Produce a structured Markdown brief with these sections:
| Section | Content |
|---|
| Stack Summary | Language, framework, build, deployment target |
| Architecture Boundaries | Ports, adapters, service boundaries, API surfaces |
| Key Files | Entrypoints, config, CI, deployment manifests |
| Conventions Detected | Naming, error handling, testing, file organization |
| Gaps / Risks | Missing tests, undocumented APIs, stale dependencies, security concerns |
| Recommended Next Steps | What to investigate further, what to plan first |
The brief is your deliverable. Send it via team_send_results.
Principles
- Scan breadth-first, then depth on areas relevant to the task
- Report what IS, not what SHOULD BE -- observations before opinions
- If the repository has an
@ai-rules: header convention, follow it
- Time-box discovery to 5 minutes. A partial brief delivered fast is more valuable than a perfect brief delivered late