tapestry
A composable Rust framework for building cross-platform AI agents and applications.
Open infrastructure for AI agents
Weaving models, tools, and platforms into capable agents
TapestryWorks builds open infrastructure and applications for cross-platform AI agents. Our stack covers the runtime, SDKs, workspace capabilities, and reference applications needed to build agents that work consistently across macOS, Windows, and Linux.
- Core language
- Rust
- First-class SDKs
- Rust & Python
- Platforms
- macOS · Windows · Linux
Projects
Composable building blocks for agent frameworks, runtimes, developer SDKs, and workspace tooling.
tapestry-runtime
The secure, cross-platform runtime for executing Tapestry agents, tools, and workflows.
tapestry-rs
Official Rust SDK for building agents, tools, and applications with Tapestry.
tapestry-py
Official Python SDK for building agents, tools, and applications with Tapestry.
tapestry-workspace
A cross-platform agent workspace file manager for scoped file operations and shell execution.
agent-workspace
A constrained workspace CLI and MCP server for agents to safely read, write, and manage files inside a project boundary.
Principles
How we design agent infrastructure that stays usable in production.
- Cross-platform by design Platform differences should be explicit and manageable, not an afterthought.
- Capabilities over implicit privilege Agent access to files, processes, and external systems should be scoped, observable, and revocable.
- Composable primitives Applications should adopt only the runtime, tools, or workflows they need.
- Developer-friendly and open Clear APIs, predictable behavior, and an ecosystem that is easy to extend.
Get involved
We welcome issues, discussions, documentation improvements, and new integrations. Start with the repository most relevant to your use case, then share feedback through GitHub.
Browse the organization