Skip to content
Agentmart
Submit a tool

composio

Composio action discovery and shell usage

Skill by vibetasking

IndexedMIT
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/vibetasking-composio

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Composio action discovery and shell usage

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/composio

What's inside

Skill name
composio
Description
Composio action discovery and shell usage
License
MIT
Instructions
52 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Composio

1000+ third-party services (Gmail, Slack, Google Sheets, etc.) through a unified CLI.

## Commands

```bash
composio gmail --help                 # list actions for a toolkit
composio gmail fetch_emails --help    # show parameters for an action
composio gmail fetch_emails --json '{"query": "from:user@example.com", "max_results": 5}'
composio slack send_message --json '{"channel": "#general", "text": "Hello"}'
```

You can also pipe JSON via stdin: `echo '{"param": "value"}' | composio <toolkit> <action>`

For a payload built from other commands' output (e.g. `jq`), or any large body, write the JSON to a file and pass `--json-file`, which reads it byte for byte, rather than quoting it inline:

```bash
jq -n --arg id "$ticket_id" '{ticket_id: $id, comment_public: false}' > payload.json
composio zendesk update_zendesk_ticket --json-file payload.json
```

`--json`/`--json-file` payloads are parsed as JSON, not Python: use lowercase `true`/`false`/`null`, not Python's `True`/`False`/`None`. If you're assembling the payload with a Python snippet, run it through `json.dumps` (or hand-write the literal) rather than interpolating Python's `repr` of a bool/`None`.

## Discovery

1. Check `<integrations>` in your system prompt. Listed services are already connected by the user, and the integration name gives you the toolkit name. Go directly to step 3.
2. Grep `system/composio/toolkits.json` for keywords (compact toolkit names and descriptions), or have a sub-agent explore it more broadly.
3. `composio <toolkit> --help`: list available actions and their descriptions for a toolkit.
4. `composio <toolkit> <action> --help`: show parameters for an action.

Get an action's exact shape before your first call to it in this run, from one of two places. When the service has its own skill (check your skill index), read the skill: a command it spells out with arguments is complete, call it as written with no `--help` round trip. For any other action, run `--help` first instead of guessing a name from what "sounds right" (e.g. a REST-API-style name).

Recipe shortcut: skip discovery when your recipe provides a known successful approach.

## Don't chain mutating calls with `&&`

When applying the same mutation across several records (e.g. updating 8 tickets), don't chain the `composio` invocations with `&&` in one shell call. `&&` stops at the first failure, so one bad field or argument on record 3 silently drops records 4-8 with no update attempted at all — and the truncated output can look like a clean run at a glance. Issue each mutating call separately (or loop with per-call status checks) so a single failure doesn't take out the rest of the batch, and you can see exactly which records still need the update.

## Tips & Tricks

- Most query actions have parameters that allow you to apply filters or paginate their results. Use them. Third-party services can have tons of data, and it's not practical to read all of it at once, as most of it won't be relevant for the task at hand. Apply strict filters, gradually relaxing them if required.

- When no parameter filters what you need, pipe the action's output through shell tools instead of reading it raw. Always `tee` the full output to a file first, so a missed match doesn't force a repeat call:

```bash
composio gmail fetch_emails --json '{"max_results": 100}' | tee emails.json | grep -i -B3 -A8 miguel
```

Action output is pretty-printed JSON: `jq` selects whole records, `grep` only matching lines. If your filter comes up empty, re-search the saved file before calling the action again.

From vibetasking/vibetasking-plugin, MIT.

More from vibetasking/vibetasking-plugin

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Oct 1, 2026 (27 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from vibetasking

Cite this page

composio skill. Agent Mart. https://agentmart.dev/skills/vibetasking-composio (last updated 2026-10-24)