TL;DR
A companion guide to the Composio CLI video: one command-line layer that lets Claude Code, OpenClaw, Codex, and other agent harnesses search, authenticate, and execute tools across 1,000+ apps.
| Resource | Description |
|---|---|
| Watch: Composio CLI with OpenClaw and Claude Code | The full 10-minute walkthrough on the DevDigest channel |
| Composio CLI | Official CLI landing page with the command reference |
| Composio CLI docs | Install, login, and the search/execute/link workflow |
| Composio dashboard | Manage connected accounts and API keys |
The video introduces the Composio universal CLI: a single command-line layer that connects AI agents to 1,000+ apps through prebuilt connectors, with OAuth and account setup handled for you. The demos build a "Hello World" Google Doc, push the latest five Hacker News stories into a Google Sheet, and then run the same kind of natural-language workflow from an OpenClaw bot over Telegram.
This post is the companion guide. Watch the video for the live demos, then use this page for the exact commands and the setup order.
If you want the broader Composio picture (SDKs, the Vercel AI SDK integration, MCP servers), read Composio 101 first. This post is narrower: it is about the CLI path and why it works so well inside agent harnesses.
The core argument in the video: a CLI is usable by both humans and agents, and LLMs are already very good at writing bash. Instead of registering dozens of MCP servers per tool, the agent gets one binary with a simple loop, and the syntax is often simpler than the equivalent MCP wiring.
It is also portable. Claude Code, Codex, OpenClaw, Cursor, VS Code, and Windsurf can all shell out to the same composio binary, so your integrations survive a change of harness. An agent can even load the usage contract on demand by running the CLI's help output, which is exactly the progressive-disclosure pattern MCP was designed around, done with plain commands.
From the official docs:
curl -fsSL https://composio.dev/install | bash
composio login
Login opens the browser and ties the CLI to your Composio account, which is where connected app accounts live.
Newsletter
Get the weekly deep dive
Tutorials on Claude Code, AI agents, and dev tools, delivered free every week.
From the archive
Jul 12, 2026 • 8 min read
Jul 12, 2026 • 9 min read
Jul 12, 2026 • 7 min read
Jul 12, 2026 • 11 min read
The CLI documents a three-step workflow for agents:
composio search "<what you want done>" finds relevant tools by natural language.composio execute <slug> -d '<params>' runs the tool slug that search returned.composio link <toolkit> connects the account in the browser, then retry the execute.The docs show the Gmail version of this loop verbatim:
composio search "summarize my unread gmail"
composio execute GMAIL_FETCH_EMAILS --get-schema
composio link gmail
composio execute GMAIL_FETCH_EMAILS -d '{ query: "is:unread newer_than:1d" }'
--get-schema prints the tool's input schema before you run it, and --dry-run validates a call without executing. Two more commands round out the surface: composio run executes inline TypeScript/JavaScript with injected helpers for multi-step workflows, and composio proxy <url> --toolkit <name> gives curl-like raw API access with Composio-managed auth.
Nothing special is required: Claude Code already has a shell. Once the CLI is installed and logged in, you can prompt something like "use the composio CLI to create a Google Sheet with the top five Hacker News stories" and the agent runs the search, execute, link loop itself. The video's Hacker News to Sheets demo is exactly this: fetch stories, then write titles, links, and points into a new sheet, with the one-time composio link googlesheets auth happening in the browser.
The video's second half wires the same workflows into an OpenClaw bot reached over Telegram. Since OpenClaw agents can run shell commands, the CLI path works there the same way, and natural-language requests in chat become scheduled or on-demand cross-app tasks without manual orchestration.
Composio also offers a hosted MCP route for OpenClaw: per composio.dev/openclaw, you add an MCP server named composio with transport type HTTP at https://connect.composio.dev/mcp, with no auth headers, since OAuth is handled automatically. Use MCP if you prefer OpenClaw's native tool registry; use the CLI if you want one portable layer across every harness you run.
Whichever path you pick, connecting an agent to your email and documents deserves a pause: run through the agent security checklist before granting scopes.
You need a Composio account. composio login authenticates the CLI through the browser, and connected app accounts (Gmail, Google Sheets, and so on) are added per toolkit with composio link.
It searches first. composio search "<task in natural language>" returns matching tool slugs, and composio execute <slug> --get-schema shows the expected input before running anything.
Not as an application runtime. The official docs say not to build production integrations on the CLI because it is in constant development with no CLI-level SLAs. It is great for agent workflows and personal automation; use the SDKs for products.
No. It is an alternative transport for the same catalog. Composio ships MCP servers too, including the hosted endpoint OpenClaw can use. The CLI wins when you want one layer across many harnesses; MCP wins when your client has native support and no shell.
Three demos: a "Hello World" Google Doc, a Google Sheet auto-populated with the latest five Hacker News stories (titles, links, points), and the same style of workflow driven from an OpenClaw bot over Telegram.
Watch the full walkthrough above, then install the CLI and run your first composio search to see what your agents can reach.
Read next
Composio is a tool infrastructure layer that connects AI agents to Gmail, GitHub, Slack, Google Calendar, and hundreds more apps - all auth handled for you. Here is how to set it up and start building real cross-app workflows.
7 min readMCP lets AI agents connect to databases, APIs, and tools. Here is what it is and how to use it in your TypeScript projects.
5 min readBefore an AI agent gets tools, files, APIs, MCP servers, or deployment access, decide what it can read, write, call, log, and roll back.
8 min readTechnical content at the intersection of AI and development. Building with AI agents, Claude Code, and modern dev tools - then showing you exactly how it works.
A hosted infinite canvas your headless AI agents drive over MCP. Any MCP-speaking agent - Claude Code, Codex, Cursor, or...
View ToolCentralized manager for MCP servers. Connect once to localhost:37373 and access all your servers through a single endpoi...
View ToolInteractive TUI dashboard that shows exactly where your Claude Code and Cursor tokens are going, in real time.
View ToolMac app for running parallel Claude Code, Codex, and Cursor agents in isolated workspaces. Watch every agent work at onc...
View ToolEvery coding agent in one window. Stop alt-tabbing between Claude, Codex, and Cursor.
View AppTurn a one-liner into a working Claude Code skill. From idea to installed in a minute.
View AppUnlock pro skills and share private collections with your team.
View AppConnect external tools and data sources via the open MCP standard.
Claude CodeThe primary command-line entry point for Claude Code sessions.
Claude CodeReal-time prompt loop with history, completions, and multiline input.
Claude Code
Composio: Connect AI Agents to 1,000+ Apps via CLI (Gmail, Google Docs/Sheets, Hacker News Workflows) Check out Composio here: http://dashboard.composio.dev/?utm_source=Youtube&utm_channel=0426&utm_

Open Design: Open-Source n8n App That Turns Any Website into a Brand Kit, Design System, HTML + Images The video introduces Open Design, an MIT-licensed full-stack template that combines AI and n8n a

Nimbalyst Demo: A Visual Workspace for Codex + Claude Code with Kanban, Plans, and AI Commits Try it: https://nimbalyst.com/ Star Repo Here: https://github.com/Nimbalyst/nimbalyst This video demos N

Composio is a tool infrastructure layer that connects AI agents to Gmail, GitHub, Slack, Google Calendar, and hundreds m...

MCP lets AI agents connect to databases, APIs, and tools. Here is what it is and how to use it in your TypeScript projec...

Before an AI agent gets tools, files, APIs, MCP servers, or deployment access, decide what it can read, write, call, log...

AI agents use LLMs to complete multi-step tasks autonomously. Here is how they work and how to build them in TypeScript.

Stop the approval-fatigue prompts without going full YOLO mode. A hands-on guide to Claude Code's permission system - se...

Databricks open-sourced Omnigent, a meta-harness that sits above individual agent CLIs so your sessions, policies, and s...

New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.