AgentSkillsCN

codebase-context

生成并维护代码库上下文文档(.context.md文件)。 实施代码库上下文规范,以实现面向AI的项目文档。在记录架构或入职时激活。

SKILL.md
--- frontmatter
name: codebase-context
description: |
  Generates and maintains codebase context documentation (.context.md files).
  Implements the Codebase Context Specification for AI-friendly project
  documentation. Activates when documenting architecture or onboarding.
allowed-tools:
  - Read
  - Write
  - Edit
  - Glob
  - Grep

Codebase Context Documentation Skill

This skill implements the Codebase Context Specification for creating AI-friendly project documentation.

When This Skill Activates

  • Starting new projects or modules
  • Documenting architecture decisions
  • Creating developer onboarding guides
  • Updating project context files
  • Generating .context.md hierarchies

Context File Formats

Markdown (.context.md)

Primary format with YAML frontmatter and markdown body.

YAML (.context.yaml)

Pure YAML for machine parsing.

JSON (.context.json)

JSON format for programmatic access.

Context Hierarchy

code
project-root/
├── .context.md           # Project-level context
├── .contextignore        # Patterns to ignore
├── .contextdocs.md       # External doc references
├── src/
│   ├── .context.md       # Source directory context
│   ├── api/
│   │   └── .context.md   # API module context
│   └── components/
│       └── .context.md   # Components context
└── tests/
    └── .context.md       # Testing context

Context File Template

yaml
---
module-name: [Module Name]
version: [X.Y.Z]
description: |
  [Comprehensive description of the module's
  purpose, functionality, and scope]

related-modules:
  - name: [Related Module]
    path: [./relative/path.md]

technologies:
  - [Technology 1]
  - [Technology 2]

conventions:
  - [Convention 1]
  - [Convention 2]

architecture:
  style: [Architecture pattern]
  components:
    - [Component 1]
    - [Component 2]
  data-flow: |
    [Description of data flow]

development:
  setup-steps:
    - [Step 1]
    - [Step 2]
  build-command: [command]
  test-command: [command]

business-requirements:
  key-features:
    - [Feature 1]
    - [Feature 2]
  target-audience: [Audience description]
  success-metrics:
    - [Metric 1]

quality-assurance:
  testing-frameworks:
    - [Framework 1]
  coverage-threshold: [percentage]
  performance-benchmarks:
    - [Benchmark 1]

deployment:
  platform: [Platform name]
  cicd-pipeline: [Pipeline description]
  staging-environment: [Staging URL/location]
  production-environment: [Prod URL/location]
---

# [Module Name]

[Extended description and documentation in markdown format]

## Architecture Overview

[Detailed architecture explanation]

## Key Components

[Component documentation]

## Usage Examples

[Code examples and usage patterns]

.contextignore Format

gitignore
# Dependencies
node_modules/
vendor/
.venv/

# Build outputs
dist/
build/
*.pyc

# IDE files
.idea/
.vscode/

# Sensitive files
.env
*.key
secrets/

.contextdocs.md Format

yaml
---
contextdocs:
  - name: API Documentation
    type: url
    link: https://docs.example.com/api
    relationship: primary
    description: Official API reference

  - name: Design System
    type: local
    path: ./docs/design-system.md
    relationship: supporting

  - name: Architecture Decision Records
    type: directory
    path: ./docs/adr/
    relationship: reference
---

Context Accumulation Rules

  1. Proximity Precedence: Closer context files override distant ones
  2. Inheritance: Child contexts inherit from parent contexts
  3. Explicit Override: Explicit declarations override inherited values
  4. Merge Strategy: Arrays merge, objects deep-merge

Generation Guidelines

  1. Start at Root: Create project-level .context.md first
  2. Descend Hierarchically: Add module-level contexts as needed
  3. Keep Current: Update contexts when architecture changes
  4. Link Related: Cross-reference related modules
  5. Document Decisions: Include architecture decision rationale

Best Practices

  1. Completeness: Fill all relevant sections
  2. Accuracy: Keep information current
  3. Clarity: Write for both humans and AI
  4. Hierarchy: Organize by module structure
  5. Maintenance: Update with code changes