capa  by infragate

Unified configuration and management for AI coding agents

Created 4 months ago
284 stars

Top 91.9% on SourcePulse

GitHubView on GitHub
Project Summary

Summary

CAPA addresses the fragmented configuration landscape for AI coding agents by providing a unified package manager. It allows users to define skills, tools, rules, and plugins in a single capabilities.yaml file, which CAPA then translates into provider-specific formats for over 35 agents. This simplifies setup, ensures consistency, and optimizes agent performance through on-demand tool exposure, benefiting developers managing complex AI agent workflows.

How It Works

The core mechanism revolves around a declarative capabilities.yaml file where users specify agent capabilities. The capa install command processes this file, resolving dependencies, generating provider-specific configurations (e.g., .cursor/rules/, .claude/agents/), and registering necessary services like MCP servers. A key design choice is on-demand tool exposure, where agents only access tools as needed, significantly reducing context window size and associated costs without compromising quality. Reproducibility is ensured via capabilities.lock and a SHA-keyed cache.

Quick Start & Requirements

  • Installation:
    • macOS/Linux: curl -LsSf https://capa.infragate.ai/install.sh | sh
    • Windows: powershell -ExecutionPolicy ByPass -c "irm https://capa.infragate.ai/install.ps1 | iex"
  • Initialization: Run capa init in your project directory to create capabilities.yaml.
  • Adding Skills: Use capa add <source> (e.g., capa add anthropics/skills@frontend-design).
  • Prerequisites: May require API keys (e.g., ${BraveApiKey}). A local HTTP server runs for the web UI and API.
  • Links: Installation scripts available at capa.infragate.ai. Full documentation, schema, and registry catalog at https://capa.infragate.ai.

Highlighted Details

  • Integrates with over 35 AI coding agents, including Cursor, Claude Code, Codex, Windsurf, and GitHub Copilot.
  • On-demand tool loading benchmarks show 19-40% cost reduction with no quality drop.
  • capa sh provides a CLI interface to execute defined MCP tools and shell commands directly.
  • Automates translation of common lifecycle hooks (e.g., beforeShell) into provider-specific configurations.
  • Supports custom and public registries for discovering and managing skills and plugins.
  • Credentials are encrypted at rest and managed securely.
  • capabilities.lock ensures deterministic builds by pinning resolved commit SHAs.

Maintenance & Community

The README does not detail specific community channels (like Discord/Slack) or notable contributors/sponsorships.

Licensing & Compatibility

  • License: MIT.
  • Compatibility: Permissive MIT license allows for broad use, including integration within commercial and closed-source projects.

Limitations & Caveats

  • Providers lacking native hook support will not have lifecycle configurations applied.
  • Non-loopback network bindings for the CAPA server require explicit authentication.
  • Functionality for certain tools may depend on correctly configured environment variables (e.g., API keys).
Health Check
Last Commit

1 day ago

Responsiveness

Inactive

Pull Requests (30d)
45
Issues (30d)
19
Star History
276 stars in the last 30 days

Explore Similar Projects

Starred by Lilian Weng Lilian Weng(Cofounder of Thinking Machines Lab), Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and
59 more.

AutoGPT by Significant-Gravitas

0.1%
185k
AI agent platform for building, deploying, and running autonomous workflows
Created 3 years ago
Updated 1 day ago
Feedback? Help us improve.