Codebase Context

Codebase Context

Purpose

Analyze the existing codebase and produce a comprehensive context briefing covering architecture, data model, patterns, and constraints. This output becomes shared context for ALL council agents — it is the foundation every other skill builds on.

Scope Constraints

• Produces read-only analysis; does not modify any project files.
• Covers architecture, data model, conventions, infrastructure, and integrations.
• Does not design new schemas or API contracts — hand off to schema-design or api-design.

Inputs

• Project root directory ($WORKSPACE)
• Any existing CLAUDE.md or project documentation
• Package manifests (package.json, pyproject.toml, Cargo.toml, etc.)
• Configuration files (tsconfig.json, next.config.js, .env.example, etc.)

Input Sanitization

No user-provided values are used in commands or file paths. All inputs are treated as read-only analysis targets.

Procedure

Progress Checklist

• [ ] Step 1: Read Project Configuration
• [ ] Step 2: Map Directory Structure
• [ ] Step 3: Identify Core Architectural Patterns
• [ ] Step 4: Map the Existing Data Model
• [ ] Step 5: Catalog Key Conventions
• [ ] Step 6: Note Infrastructure Constraints
• [ ] Step 7: Identify Integration Points

Step 1: Read Project Configuration

Read CLAUDE.md, package.json (or equivalent), tsconfig, and any config files at the project root. Extract framework, language version, key dependencies, and build tooling.

Step 2: Map Directory Structure

List top-level directories and identify their purposes. Recurse one level into key directories (src/, app/, lib/, components/, etc.) to understand the organizational pattern.

Step 3: Identify Core Architectural Patterns

Determine the routing approach (file-based, config-based), data fetching strategy (SSR, SSG, CSR, RSC), state management solution, and component patterns (atomic, feature-based, etc.).

Step 4: Map the Existing Data Model

Read database schema files, migration history, or ORM models. List tables, their relationships, key entities, and access control policies (e.g., RLS policies for Supabase).

Step 5: Catalog Key Conventions

Identify naming conventions (camelCase vs snake_case, file naming), file structure patterns, import conventions (barrel exports, path aliases), and error handling patterns used throughout the codebase.

Step 6: Note Infrastructure Constraints

Document hosting platform, database type and version, auth provider, CDN configuration, edge function usage, environment variable patterns, and deployment pipeline.

Step 7: Identify Integration Points

Catalog external APIs, webhooks, third-party services, and any inter-service communication patterns. Note authentication methods for each integration.

> Compaction resilience: If context was lost during a long session, re-read the Inputs section to reconstruct what system is being analyzed, check the Progress Checklist for completed steps, then resume from the earliest incomplete step.

Handoff

• If the analysis reveals schema issues, inconsistencies, or new entity requirements, recommend loading architect/schema-design.
• If the analysis reveals API surface gaps, endpoint inconsistencies, or missing contracts, recommend loading architect/api-design.

Output Format

# Context Briefing

## Tech Stack
- **Framework**: [name + version]
- **Language**: [name + version]
- **Database**: [type + provider]
- **Auth**: [provider + method]
- **Hosting**: [platform]
- **Key Dependencies**: [list with versions]

## Directory Structure

[tree output with annotations]

## Architectural Patterns
- **Routing**: [approach]
- **Data Fetching**: [strategy]
- **State Management**: [solution]
- **Component Pattern**: [style]

## Data Model
| Entity | Key Fields | Relationships |
|--------|-----------|---------------|
| ...    | ...       | ...           |

[ASCII relationship diagram if complex]

## Conventions
1. [Convention with example]
2. [Convention with example]
3. [Convention with example]

## Infrastructure Constraints
- [Constraint and implication]
- [Constraint and implication]

## Integration Points
| Service | Purpose | Auth Method |
|---------|---------|-------------|
| ...     | ...     | ...         |

## Warnings / Tech Debt
- [Known issue or constraint that affects design decisions]

Quality Checks

• [ ] No hardcoded file paths — all paths are relative to $WORKSPACE
• [ ] All 7 process areas are covered in the output
• [ ] At least 3 conventions are identified and documented with examples
• [ ] Tech stack versions are specified (not just names)
• [ ] Data model section includes relationships, not just entity lists
• [ ] Integration points include auth methods
• [ ] Warnings section flags any constraints that would affect new feature design

Evolution Notes

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: dtsong
• Source: dtsong/my-claude-setup
• License: MIT

Install and usage instructions live in the source repository linked above.