puppyone  by puppyone-ai

Git-powered context management for AI agents

Created 1 year ago
862 stars

Top 40.9% on SourcePulse

GitHubView on GitHub
Project Summary

Summary

Puppyone addresses the challenge of managing context for AI agents by providing a Git-native Context Drive. It offers version control, file-level scoped permissions, and audit trails, enabling secure and organized data access for AI applications. This system is designed for AI developers and power users seeking robust context management solutions.

How It Works

Puppyone implements a Git-based approach for context hosting, leveraging Git's versioning capabilities for file history and rollback. Its core innovation lies in File-Level Security (FLS), which enforces granular, per-agent permissions at the filesystem layer, ensuring agents only access authorized data. A central "Context Drive" serves multiple "Access Points," allowing different AI agents and tools to interact via various protocols like MCP, REST API, CLI, or local sync, each with tailored security scopes.

Quick Start & Requirements

  • Installation:
    • Cloud: Create an account at puppyone.ai.
    • Self-Hosted: Requires Docker. Clone the repository, navigate to the docker directory, copy .env.example to .env, and run docker compose up -d. First startup may take 1-2 minutes.
  • CLI: Install globally via npm install -g puppyone and authenticate using puppyone auth login.
  • Prerequisites: Docker for self-hosting. Optional: ANTHROPIC_API_KEY for agent chat, provider credentials for OAuth connectors (GitHub, Gmail, Google Drive).
  • Links: Cloud: https://puppyone.ai/, Repository: https://github.com/puppyone-ai/puppyone.git.

Highlighted Details

  • Extensive Connectors: Integrates with SaaS tools (GitHub, Gmail), Google Workspace (Drive, Docs, Sheets, Calendar, Search Console), web pages, local filesystems, and Supabase databases.
  • Agent-Centric Security: Features File-Level Security (FLS) for granular access control, version history with rollback, folder-level snapshots, and comprehensive audit logs for all operations.
  • Collaborative Workflow: Supports checkout/commit semantics with locking and conflict detection for concurrent agent edits.
  • Flexible Access: Provides multiple access methods including MCP endpoints (for Cursor, Claude Desktop), secure sandboxes (Docker/E2B), a REST API, a CLI, and bidirectional local folder sync via OpenClaw.

Maintenance & Community

The provided README does not detail specific maintenance schedules, notable contributors, sponsorships, or community channels like Discord or Slack. Development guidelines are referenced via CONTRIBUTING.md.

Licensing & Compatibility

Puppyone is licensed under the Apache License 2.0, permitting broad use including commercial products and managed services, provided copyright notices are preserved and modifications are documented. However, "Puppyone" and its logo are trademarks and not included in the Apache 2.0 license; derivative works must use different branding.

Limitations & Caveats

For local self-hosted deployments, the default sandbox configuration shares the host's Docker daemon for convenience, which may pose security risks in multi-tenant or remote environments; using SANDBOX_TYPE=e2b with an E2B API key is recommended for enhanced security in such scenarios. Optional features like agent chat and OAuth connectors require additional configuration and API keys.

Health Check
Last Commit

3 days ago

Responsiveness

Inactive

Pull Requests (30d)
33
Issues (30d)
0
Star History
657 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.