What problem it solves

Coding agents commonly pause for human approval before risky steps, which makes unattended repository work difficult to run in CI, daemons, and scheduled jobs. Teams also need consistent policies, tools, and supervision across different execution environments without maintaining separate framework integrations.

Key features

• Repository tooling through MCP — Give an LLM tools for reading and editing files, running shells, searching code, and delegating to sub-agents through MCP.

  Source: README

• Policy as code — Define deterministic pre-call guards, post-result hooks, and post-turn validators in TOML and scripts. These policies can enforce restrictions without recurring approval prompts.

  Source: README

• Supervisor and session learning — A second model maintains the plan, checks completion, and condenses tool output. Lessons extracted from completed work can be recalled by later sessions for the same project and role.

  Source: README

• Specialists and configurable roles — Install packaged specialists that include prompts, models, MCP servers, and permissions. Roles, workflows, tools, and models can also be configured for different tasks.

  Source: README

• Multiple execution surfaces — Use the same harness through an interactive CLI, non-interactive output, a daemon, a WebSocket server, or ACP editor integration.

  Source: README

Quick start

Copied verbatim from the project's README.

curl -fsSL https://raw.githubusercontent.com/muvon/octomind/master/install.sh | bash

octomind login

cd ~/your/project
octomind run developer:general

Who it may suit

Inference: This project may suit developers who need coding agents in interactive terminals, CI automation, background daemons, or editor integrations. It may also suit teams that want repository policies and agent configuration represented as code.

Topics to discuss

• Which deterministic guardrails should be included in a standard repository policy?

• How should delegated agents receive project context and supervisor guidance?

• Which specialist agent configuration would be most useful to publish first?

Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.

Read the original README https://github.com/Muvon/octomind#readme

ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.