Devotd Github
NewAI agents

composiohq/composio — SDK of app tools for AI agents

composiohq/composio is an SDK monorepo that helps AI agents discover and authenticate external app tools and use them for tasks.

Repositorycomposiohq/composio ↗

composiohq/composio is an SDK monorepo that helps AI agents discover and authenticate external app tools and use them for tasks.

GitHub description: Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.

stars
★ 30.7K
Forks
4.9K
Language
TypeScript
License
MIT
Latest release
@composio/anthropic@0.11.6
Read original source ↗View on GitHub ↗

composiohq/composio is an SDK monorepo that helps AI agents perform tasks using tools from external apps. The repository description lists as its main features toolkits of more than 1000 apps, tool search, context management, authentication, and a sandbox workbench. According to the README, it also provides pre-authenticated toolkits, per-user sessions, and triggers. The latest release information provided is `@composio/anthropic@0.11.6`, and there is no record of what changed in this version.

Sessions provide meta tools that find the tools to use during a run and authenticate and execute them, and the hosted MCP endpoint is also exposed. The repository contains TypeScript and Python SDKs, the `composio` CLI, and Provider adapters for multiple agent frameworks. Therefore, the core of Composio lies not in individual app features but in the foundation that connects multiple app tools to agents and makes them usable. The repository language is TypeScript and the license is MIT. The star count recorded in FACTS is 30742.

However, the provided records contain no performance measurements or comparisons with other tool-integration projects. The version of the latest release is confirmed, but the changelog is not given, so this material alone cannot be used to judge what features the current release has added. The momentum score is 5.5 points, and the technical/quality score is not recorded. Therefore, the scope of features can be confirmed, but there is no evidence in these records backing claims of superiority in performance or implementation quality.