copilot-agents
Custom agents built for GitHub Copilot to automate agentic coding workflows.
Available Agents
🔍 Codebase Analysis Agent
Agent Name: codebase-analysis
Purpose: A specialist at understanding HOW code works - analyzes implementation details, traces data flow, and explains technical workings with precise file:line references.
Use Cases:
- Understanding how a specific function or module works
- Tracing data flow through the application
- Analyzing implementation patterns and architecture
- Getting detailed explanations of complex code with exact references
- Understanding control flow and execution paths
- Mapping dependencies between components
How to Use: Invoke the agent in GitHub Copilot by mentioning it in your chat:
@codebase-analysis How does the authentication flow work in this codebase?
@codebase-analysis Trace the data flow from the API endpoint to the database
@codebase-analysis Explain how the payment processing is implemented
What You'll Get:
- Precise file:line references for every code element
- Step-by-step data flow traces
- Detailed implementation explanations
- Control flow analysis
- Technical insights with proper context
Key Features:
- ✅ Provides exact file:line references (e.g.,
src/auth/login.js:45-67) - ✅ Traces data transformations step-by-step
- ✅ Explains implementation patterns and design choices
- ✅ Documents dependencies and component interactions
- ✅ Includes relevant code snippets with line numbers
- ✅ Analyzes error handling and edge cases
Configuration: .github/agents/codebase-analysis.yml
---
📁 Code Finder Agent
Agent Name: code-finder
Purpose: A specialist at finding WHERE code lives in a codebase - locates relevant files and organizes them by purpose, without analyzing their contents.
Use Cases:
- Quickly locating files related to a specific feature or topic
- Understanding project file organization and structure
- Finding all files of a specific type (tests, configs, docs, types)
- Discovering where to add new code for a feature
- Mapping out file clusters and related code locations
- Getting an inventory of files by category
How to Use: Invoke the agent in GitHub Copilot by mentioning it in your chat:
@code-finder Find all files related to authentication
@code-finder Locate payment processing code
@code-finder Show me all test files for the API layer
@code-finder Where should I add new shopping cart code?
What You'll Get:
- Organized file lists by category (implementation, tests, config, docs, types, examples)
- Full paths from repository root
- Directory clusters showing related files
- File organization patterns
- Recommendations for where to add new files
Key Features:
- ✅ Fast file location using grep, find, and glob patterns
- ✅ Categorizes files by purpose (implementation, tests, config, docs, types, examples)
- ✅ Shows directory organization and file clusters
- ✅ Provides full paths and file counts
- ✅ Identifies naming patterns and conventions
- ✅ Notes common locations for different file types
Configuration: .github/agents/code-finder.yml
---
🔎 Pattern Finder Agent
Agent Name: pattern-finder
Purpose: A specialist at finding code patterns and examples in the codebase - locates similar implementations that can serve as templates or inspiration for new work.
Use Cases:
- Finding similar implementations to use as templates
- Discovering how specific patterns are implemented
- Locating usage examples for a particular approach
- Understanding coding conventions through examples
- Finding test patterns for similar features
- Exploring multiple variations of the same pattern
How to Use: Invoke the agent in GitHub Copilot by mentioning it in your chat:
@pattern-finder Show me pagination patterns in the codebase
@pattern-finder Find error handling patterns in API routes
@pattern-finder How is authentication implemented? Show me examples
@pattern-finder Find state management patterns in React components
What You'll Get:
- Multiple pattern examples with code snippets
- File:line references for each pattern
- Key characteristics of each pattern
- Testing patterns for the implementation
- Information about where patterns are used
- Related utilities and helper files
Key Features:
- ✅ Shows existing patterns without judgment or critique
- ✅ Provides multiple variations of the same pattern
- ✅ Includes complete code examples with context
- ✅ Documents test patterns alongside implementation patterns
- ✅ Notes where each pattern is used in the codebase
- ✅ Focuses on "what exists" rather than "what should be"
- ✅ Catalogs patterns like a reference library
Configuration: .github/agents/pattern-finder.yml
---
Superpowers Integration
This repository includes the Superpowers system integration, which provides additional skills and capabilities for AI coding agents. The superpowers system is located in .codex/superpowers/ and includes skills for:
- Test-driven development
- Systematic debugging
- Code review workflows
- Plan writing and execution
- Root cause tracing
- Git worktree management
- And many more...
Using Superpowers
The superpowers system is configured in .codex/AGENTS.md and can be activated by running:
.codex/superpowers/.codex/superpowers-codex bootstrap
You can also add your own custom skills to .codex/skills/.
For more information, see the Superpowers documentation.
---
Setup
These custom agents are automatically available when this repository is configured with GitHub Copilot. The agent configurations are stored in the .github/agents/ directory.
Contributing
To add new agents: 1. Create a new YAML file in .github/agents/ 2. Define the agent's name, description, and instructions 3. Update this README with documentation 4. Test the agent in GitHub Copilot
Agent Structure
Each agent is defined with:
- name: Unique identifier for the agent
- description: Brief summary of the agent's purpose
- instructions: Detailed prompt that guides the agent's behavior
License
See repository license for details.











