arbazkhan971/onboard

Activate When

• /godmode:onboard, "explain this codebase"
• "architecture overview", "how does this work?"
• New team member needs project orientation

Workflow

1. Project Discovery

ls package.json Cargo.toml go.mod pyproject.toml \
  pom.xml Gemfile *.csproj 2>/dev/null
find . -name "*.ts" -o -name "*.py" -o -name "*.go" \
  | grep -v node_modules | wc -l
git log --oneline -20
git log --since="30 days ago" --diff-filter=M \
  --pretty=format:"%h %s" | head -10

Name: <from manifest>
Type: web app|API|CLI|library|monorepo|microservice
Language: <primary>
Framework: <detected>
Build: <webpack|vite|tsc|make|cargo>
Tests: <jest|pytest|go test>
LOC: <approximate>

2. Architecture Walkthrough

Layer map:
| Layer | Responsibility |
| Routes | URL mapping, response formatting |
| Middleware | Auth, rate limiting, logging |
| Controllers | Validation, orchestration |
| Services | Business logic, domain rules |
| Repositories | Data access, caching |
| Models | Data shapes, relationships |

Data flow:
Client -> LB -> Middleware -> Router -> Controller
  -> Service -> Repository -> Database

3. Key Files (max 10)

Read in this order:
1. README.md — setup instructions
2. <entry-point> — bootstrap (main.ts, app.py)
3. <router> — all endpoints
4. <config> — environment, feature flags
5. <core-service> — main business logic
6. <data-model> — primary schema
7. <middleware> — request pipeline
8. <test-example> — testing patterns

IF README missing: note as gap, generate from code. IF > 500 files: focus on 20 most-modified (git log).

4. Naming Conventions

Files: PascalCase (components), camelCase (utils)
Variables: camelCase locals, SCREAMING_SNAKE constants
Functions: camelCase, verb prefix (get, create, handle)
Types: PascalCase, I-prefix for interfaces (if TS)
Tests: <name>.test.ts / <name>.spec.ts

5. Dependency Graph

Internal: routes->controllers->services->repositories
  Shared: utils, config, logger, types
  Circular deps: <none | list with file paths>
External:
  Critical: <pkg — purpose>
  Dev-only: <pkg — purpose>

6. Code Tour (5+ stops)

Stop 1: Entry Point — where everything starts
Stop 2: Configuration — env handling
Stop 3: Core Route — a typical request lifecycle
Stop 4: Data Model — primary entities
Stop 5: Test File — testing patterns and conventions

7. Onboarding Report

Project: <type> with <framework>
Size: ~<N> files, ~<N>K lines
Architecture: <monolith|microservices|monorepo>
Pattern: <MVC|Clean|Hexagonal>
Key files: <N> identified
Setup: <N> steps to first run

<!-- tier-3 -->

Quality Targets

• Target: <15min from clone to running dev server
• Target: <5 manual steps in setup process
• Target: 100% of environment variables documented

Hard Rules

1. NEVER claim architecture without reading code.
2. NEVER list > 10 key files (80/20 rule).
3. NEVER skip git history (most-modified files).
4. NEVER assume REST — detect actual protocol.
5. ALWAYS read README first if it exists.
6. ALWAYS identify entry point before anything else.
7. ALWAYS include test file in code tour.

TSV Logging

Append .godmode/onboard-results.tsv:

timestamp	repo	files_analyzed	entry_points	status

Keep/Discard

KEEP if: all claims verified against source code
  AND output populated with real data.
DISCARD if: unverified claims OR placeholder text.

Stop Conditions

STOP when FIRST of:
  - Architecture identified + key files documented
  - Code tour has 5+ stops
  - Dev setup verified (build + test from clean)

Autonomous Operation

On failure: git reset --hard HEAD~1. Never pause.

Error Recovery

| Failure | Action | |--|--| | Build fails | Check system deps, env vars, platform | | Architecture ambiguous | Check manifest, entry points, infra | | Monorepo detected | Repeat architecture per workspace |