Back to directory

MS-Agent

A modular, extensible open-source agent framework for long-running tasks, combining models, tools, skills and sub-agents behind a customizable harness that manages planning, context, permissions and execution feedback.

Visit website (opens in a new tab)ms-agent-en.readthedocs.io
MS-Agent

MS-Agent is an agent framework for building a productivity assistant tailored to complex, long-running tasks, combining models, tools, skills and sub-agents behind a customizable harness that handles planning, context, permissions and execution feedback, with project memory and autonomous scheduling to keep tasks moving forward. A shared Python SDK backs a CLI, a TUI and a WebUI, so the same agent logic and extensions can be reused across terminals, browser workspaces and business applications.

Its Agentic Insight deep-research system has placed near the top of the DeepResearch Bench leaderboard, and recent releases added context compression with automatic pruning and summarization, a knowledge-search integration for retrieval over local codebases and documentation, and multimodal image and video input. An Agent Hub manages agent workspace files locally and in remote repositories, including uploads, downloads, background sync, conversion between agent frameworks, and backup and restore.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.