About this project

TrueForge is an open-source agent harness that serves as the runtime layer for transforming an LLM into a functional agent. It manages the entire agent execution loop, including model calls, MCP tools, skills, sandboxing, approvals, context management, and session state. The platform exposes these capabilities through three interfaces: a chat UI, an HTTP API with a TypeScript SDK, and an embeddable UI SDK. Key features include: - **Initial setup from catalogs**: Configure models, MCP servers, skills, and sandboxes once; agents can then select from the connected resources. Presets come from customizable YAML catalogs. - **Any model provider**: Supports OpenAI, Anthropic, Google Gemini, and other catalog providers, or any OpenAI-compatible endpoint. - **MCP tools**: Remote MCP servers with header auth or OAuth, including in-chat authorization. - **Skills**: Git-backed SKILL.md instruction packs loaded on demand in the sandbox. - **Sandbox as a tool**: Isolated code/file execution (Daytona currently; more providers planned), provisioned only when needed. Secrets remain in the harness. - **Human checkpoints**: Tool approval, ask-user-questions, and Generative UI in chat. - **Context engineering**: Subagents, deferred tool loading, Code Mode, large-result offloading, and compaction. - **Chat UI + SDK**: Use the bundled UI, automate with @truefoundry/trueforge-sdk, or embed @truefoundry/trueforge-ui. TrueForge scales from local mode (single process, SQLite) to hosted mode (Postgres + Redis, Docker Compose, Helm, or Railway). Quickstart is available via `npx @truefoundry/trueforge@latest`. The project is MIT-licensed and welcomes contributions.