composio-automation
by ComposioHQcomposio-automation helps agents run Composio workflows through Rube MCP by discovering current tool schemas, checking the Composio connection, then executing and verifying tasks.
This skill scores 68/100, which means it is acceptable for listing but should be presented as a lightweight connector/pattern skill rather than a complete workflow pack. Directory users get enough information to know it is for Composio operations via Rube MCP and how an agent should start, but they should expect the agent to rely on live tool discovery for most task-specific details.
- Clear trigger and dependency framing: it explicitly targets Composio automation through Rube MCP and declares `requires: mcp: [rube]`.
- Setup and prerequisite guidance identifies the required MCP endpoint, connection check, and ACTIVE Composio toolkit connection before use.
- Strong schema-discovery instruction: it repeatedly tells agents to call `RUBE_SEARCH_TOOLS` first so they use current tool schemas instead of guessing.
- Operational detail is mostly a generic discovery/check/execute pattern rather than concrete Composio task workflows or tested examples.
- The skill depends entirely on Rube MCP availability and an active Composio connection; there are no support scripts, references, or local assets beyond SKILL.md.
Overview of composio-automation skill
What composio-automation does
composio-automation is a Claude skill for running Composio-related workflows through Rube MCP. Its main value is not a fixed script; it teaches the agent to discover the current Composio tool schemas first, verify the user’s Composio connection, and then execute the right Rube MCP tool sequence for the requested automation task.
Use this skill when you want an agent to interact with Composio operations rather than merely explain Composio concepts.
Best fit for Workflow Automation users
The composio-automation skill is best for users building or operating Workflow Automation around Composio toolkits, integrations, connected accounts, or automation tasks exposed through Rube. It is especially useful when tool schemas may change, because the skill explicitly instructs the agent to call RUBE_SEARCH_TOOLS before attempting execution.
It fits teams that already use Claude with MCP tools and want a repeatable pattern for “discover, authenticate, execute, verify” instead of ad hoc prompting.
Key adoption requirements
Before installing, confirm your client can use MCP servers and that Rube MCP is available. The skill expects:
- Rube MCP connected with
RUBE_SEARCH_TOOLSavailable - A Composio connection managed through
RUBE_MANAGE_CONNECTIONS - Toolkit name
composio - Connection status
ACTIVEbefore workflow execution
If your environment cannot expose MCP tools to the agent, this skill will be limited to guidance only.
What makes it different from a generic prompt
A generic prompt may ask the model to “use Composio,” but it can easily guess stale tool names or invent parameters. composio-automation centers tool discovery: search first, inspect returned schemas, check connection state, then run the selected tool. That sequence reduces schema mismatch, authentication surprises, and failed automation runs.
How to Use composio-automation skill
composio-automation install context
Install the skill from the ComposioHQ skill collection:
npx skills add ComposioHQ/awesome-claude-skills --skill composio-automation
Then add Rube MCP to your MCP-capable client using:
https://rube.app/mcp
The upstream skill does not include extra scripts, references, or helper assets, so start with SKILL.md. That file is the operating procedure: prerequisites, setup, tool discovery, and the core workflow pattern.
Inputs the skill needs from you
For good composio-automation usage, give the agent a concrete task and the operational context needed to search tools accurately. Include:
- The Composio task you want completed
- The target toolkit or integration if known
- Whether the Composio connection already exists
- Any entity names, IDs, accounts, projects, or filters involved
- The expected output, such as “create,” “list,” “update,” “verify,” or “summarize results”
- Safety constraints, such as “do not delete,” “dry run first,” or “ask before modifying”
Weak prompt: “Automate Composio.”
Stronger prompt: “Use the composio-automation skill to find the current Rube MCP tools for Composio. Check whether my composio connection is ACTIVE, then list available Composio toolkits connected to my account. Do not modify anything.”
Recommended workflow pattern
A reliable composio-automation guide should follow this order:
- Call
RUBE_SEARCH_TOOLSfor the specific use case, not a vague generic query. - Reuse the returned session ID for follow-up discovery or execution.
- Call
RUBE_MANAGE_CONNECTIONSfor toolkitcomposio. - If the connection is not
ACTIVE, follow the returned authentication link. - Review the returned tool slug, schema, required fields, and pitfalls.
- Execute only after the schema is known.
- Verify the result and report what changed.
This matters because Rube can return current tool schemas and execution plans. Skipping discovery is the fastest path to invalid arguments.
Practical prompt patterns
For read-only work, say so explicitly:
“Use composio-automation to search current Composio tools via Rube MCP, confirm my connection, and retrieve the requested data. This is read-only; do not create, update, or delete resources.”
For state-changing work, require confirmation:
“Discover the correct Composio tool schema first. Prepare the exact tool call for creating the automation, explain the fields, and wait for my approval before execution.”
For troubleshooting:
“Use RUBE_SEARCH_TOOLS to identify the relevant Composio operation and known pitfalls. Check connection status with RUBE_MANAGE_CONNECTIONS, then explain whether the failure is authentication, missing fields, unavailable tool, or execution error.”
composio-automation skill FAQ
Is composio-automation suitable for beginners?
Yes, if the beginner is using an MCP-capable client and can follow an authentication link. The skill’s workflow is simple, but the surrounding ecosystem is not purely plug-and-play: users must understand that the agent is calling Rube MCP tools and that Composio access depends on an active connection.
When should I not use this skill?
Do not use composio-automation if you only need documentation, conceptual advice, or static code examples. It is also a poor fit when your client cannot connect to Rube MCP, when organizational policy blocks external MCP servers, or when you cannot authorize a Composio connection.
For non-execution research, a normal prompt plus Composio documentation may be enough.
Does it replace Composio documentation?
No. The skill points to Composio toolkit documentation and uses Rube MCP discovery for current schemas. It is an execution workflow, not a complete API reference. For edge cases, permissions, billing, or toolkit-specific behavior, read the Composio docs alongside the tool schema returned by RUBE_SEARCH_TOOLS.
Why is tool search mandatory?
Tool search is mandatory because the skill depends on live schemas. MCP tools may expose specific slugs, required fields, and constraints that differ from memory or examples. RUBE_SEARCH_TOOLS gives the agent the current execution surface before it commits to a call, which is the core reliability benefit of the composio-automation skill.
How to Improve composio-automation skill
Improve composio-automation prompts
The best way to improve composio-automation results is to replace broad goals with operational requests. Name the action, scope, connection expectation, and confirmation policy.
Instead of: “Set up a Composio workflow.”
Use: “Search current Rube MCP tools for Composio operations related to GitHub issue automation. Confirm the composio connection is ACTIVE. If a create or update action is needed, show the proposed tool call and wait for approval.”
This helps the agent choose a precise use_case for discovery and prevents accidental writes.
Avoid common failure modes
Common failures include skipping RUBE_SEARCH_TOOLS, assuming the connection is active, using stale parameter names, or treating an authentication problem as a tool problem. If a run fails, ask the agent to classify the issue:
- Discovery failure
- Connection inactive
- Missing required field
- Permission or account limitation
- Tool execution error
- Ambiguous user intent
This keeps iteration focused instead of retrying the same bad call.
Iterate after the first output
After the first run, improve quality by asking for the returned tool slug, required fields, omitted optional fields, and any warnings from Rube. For state-changing workflows, request a preflight summary before execution and a post-run verification after execution.
A strong follow-up is: “Based on the discovered schema, identify required inputs I have not provided yet and ask only for those. Do not execute until all required fields are known.”
Extend the skill for team use
Teams can improve the skill by adding local notes outside the upstream file: approved Composio use cases, naming conventions, allowed write operations, rollback expectations, and examples of safe prompts. Because the repository path currently exposes only SKILL.md, team-specific guardrails should be documented in your own workspace or agent instructions rather than assumed from the package.