From adlc-team
Systematically map an existing codebase — stack detection, architecture, domain discovery, test coverage, tech debt. Trigger: 'explore codebase', 'onboard', Case C/D repo detection. Shared across all roles.
How this skill is triggered — by the user, by Claude, or both
Slash command
/adlc-team:shared-exploreThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
<context>
Read configuration files to identify stack:
package.json, tsconfig.json → Node.js/TypeScriptgo.mod → Gopyproject.toml, requirements.txt → PythonCargo.toml → Rustpom.xml, build.gradle → Java/KotlinDockerfile, docker-compose.yml → containerizationREADME.md, CLAUDE.md — any existing docsRecord: language, framework, build tool, test framework, linter, CI/CD.
# Directory structure (depth 3)
find . -type d -not -path '*/node_modules/*' -not -path '*/.git/*' -not -path '*/vendor/*' -not -path '*/__pycache__/*' | head -50
# Entry points
# Look for main files, index files, app files, server files
Identify:
Scan for domain-specific types, interfaces, constants:
# Find type definitions, interfaces, models
grep -r "type\|interface\|class\|model\|schema" --include="*.ts" --include="*.go" --include="*.py" -l
Identify:
# Find test files
find . -name "*test*" -o -name "*spec*" | grep -v node_modules | grep -v vendor
# Count test files vs production files
Identify well-tested vs untested areas.
Look for:
git log --oneline -20Write to .sdlc/exploration-report.md:
# Codebase Exploration Report
## Stack
- **Language:** [X]
- **Framework:** [X]
- **Database:** [X]
- **Test framework:** [X]
- **Build/lint:** [X]
## Architecture
[Pattern, entry points, key directories]
## Domain Concepts
[Key entities, relationships, business rules]
[Seed terms for domain-terms.md]
## Test Coverage
- Test files: [N]
- Production files: [N]
- Well-tested areas: [list]
- Gaps: [list]
## Code Health
- TODOs: [N]
- Recent activity: [pattern]
- Key concerns: [list]
## Recommended First Steps
1. [Most impactful action]
2. [Second action]
3. [Third action]
Also generate initial scaffold files if .sdlc/ doesn't exist:
.sdlc/domain-context.md (from exploration findings).sdlc/domain-terms.md (seed from domain analysis).sdlc/verification.yml (from detected stack)This is READ-ONLY. No code modifications, no destructive commands, no database operations. If you find secrets, flag them without including values in the report.
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub obacker/claude-code-plugins --plugin adlc-team