composio-search-automation
by ComposioHQcomposio-search-automation helps Claude run Composio Search workflows through Rube MCP by discovering current tool schemas, checking the composio_search connection, and executing with fewer assumptions.
This skill scores 66/100, which makes it an acceptable but limited listing candidate. Directory users get enough information to understand that it is for Composio Search automation through Rube MCP and how to start safely, but the install decision is weakened by the lack of concrete task examples, support files, and operation-specific workflows.
- Valid frontmatter declares the required MCP dependency (`rube`) and clearly identifies the trigger area: automating Composio Search tasks via Rube MCP.
- Prerequisites and setup steps are explicit, including verifying `RUBE_SEARCH_TOOLS`, managing the `composio_search` connection, and confirming ACTIVE status before use.
- The skill gives a repeatable tool-discovery pattern and emphasizes fetching current schemas before execution, reducing guesswork compared with a generic prompt.
- No support files, scripts, examples, or references beyond SKILL.md, so adoption depends entirely on the short markdown instructions and live Rube tool discovery.
- The workflow is schema-dependent and tells agents to call RUBE_SEARCH_TOOLS first, which is correct but means the skill provides limited concrete operation-specific guidance on its own.
Overview of composio-search-automation skill
What composio-search-automation does
composio-search-automation is a Claude skill for running Composio Search toolkit workflows through Rube MCP. Its main value is not a fixed search command; it teaches the agent to discover the current Composio Search tool schemas first, verify the connection, then execute the right tool calls with fewer assumptions.
This matters because Composio/Rube tool names, inputs, and execution plans can change. The skill explicitly prioritizes RUBE_SEARCH_TOOLS before action, so the agent works from live schemas instead of stale prompt memory.
Best-fit users and workflows
Use the composio-search-automation skill when you want an AI agent to operate against Composio Search through Rube MCP, especially for workflow automation where the exact available tools must be discovered at runtime. It fits users who already use Composio, are comfortable authorizing MCP tool access, and want Claude to handle tool discovery, connection checks, and step-by-step execution.
It is most useful for tasks like finding available Composio Search operations, preparing a valid execution plan, checking required fields, and running search-related automation after the toolkit connection is active.
Key adoption requirement
The skill depends on Rube MCP. Your client must have https://rube.app/mcp configured as an MCP server, and RUBE_SEARCH_TOOLS must be available. You also need an active Composio Search connection via RUBE_MANAGE_CONNECTIONS using the composio_search toolkit.
If you cannot enable MCP tools, this skill will not provide much more than written guidance. Its practical value comes from allowing the agent to call Rube MCP tools directly.
How to Use composio-search-automation skill
composio-search-automation install and setup path
Install the skill from the ComposioHQ skill collection:
npx skills add ComposioHQ/awesome-claude-skills --skill composio-search-automation
Then configure Rube MCP in your AI client by adding https://rube.app/mcp as an MCP server. In the repository, start with composio-skills/composio-search-automation/SKILL.md; there are no companion scripts or reference folders in this skill, so the main file is the source of truth.
Before asking for an actual workflow, verify:
RUBE_SEARCH_TOOLSresponds.RUBE_MANAGE_CONNECTIONScan check toolkitcomposio_search.- The Composio Search connection status is
ACTIVE.
Required inputs for reliable usage
A weak request is: “Use Composio Search.”
A stronger request gives the agent enough context to discover the correct tools and avoid guessing:
“Use the composio-search-automation skill through Rube MCP. First call RUBE_SEARCH_TOOLS for the use case: find Composio Search tools that can search indexed company knowledge. Check the composio_search connection with RUBE_MANAGE_CONNECTIONS. If active, propose the execution plan and list required fields before running any action.”
Good inputs include:
- The exact search goal or business object.
- Whether you want discovery only, execution, or both.
- Any known fields, filters, identifiers, or expected result format.
- Whether the agent should pause before taking action.
Practical workflow for first run
A safe first-run sequence is:
- Ask the agent to invoke the composio-search-automation skill.
- Require
RUBE_SEARCH_TOOLSbefore any execution. - Ask it to summarize returned tool slugs, schemas, and pitfalls.
- Check the Composio Search connection with
RUBE_MANAGE_CONNECTIONS. - If inactive, complete the returned auth flow.
- Re-run discovery after activation.
- Execute only after the agent has mapped your goal to required tool inputs.
This pattern is slower than a direct prompt, but it is more robust for Workflow Automation because the agent bases actions on current schemas.
Prompt pattern that improves output quality
Use prompts that separate discovery, planning, and execution:
“Using composio-search-automation, discover the current Composio Search tools for [task]. Do not assume field names. After RUBE_SEARCH_TOOLS, show me the matching tool slug, required inputs, optional filters, and any pitfalls. Then ask for missing values before execution.”
This works better than a generic instruction because it matches the skill’s core design: search tools first, connect second, execute last.
composio-search-automation skill FAQ
Is composio-search-automation only for Composio users?
Yes. The composio-search-automation skill is specifically built around Composio Search via Rube MCP. If you are not using Composio Search or cannot authorize the composio_search toolkit connection, the skill is probably not the right install.
How is this better than an ordinary Claude prompt?
A normal prompt may invent tool names or rely on outdated assumptions. This skill tells the agent to call RUBE_SEARCH_TOOLS first, retrieve current schemas, and use RUBE_MANAGE_CONNECTIONS to confirm access. That makes it better for operational automation where valid tool inputs matter.
Is it beginner-friendly?
It is beginner-friendly only if your MCP environment is already working. The repository content is compact and direct, but it assumes you understand MCP tool calling, connection states, and authorization links. New users should first confirm Rube MCP is visible in their client before installing the skill.
When should I not use this skill?
Do not use it for general web search, non-Composio search APIs, or offline planning where no MCP tools can be called. Also avoid it when you need a fully packaged script or local CLI; this skill is an agent workflow instruction, not a standalone automation program.
How to Improve composio-search-automation skill
Improve composio-search-automation results with better task framing
The most important improvement is giving the agent a precise use case before discovery. Instead of “search my data,” describe the object, action, and constraints: “Find recent matching records in Composio Search for customer support articles about OAuth connection failures, return title, URL, and confidence if supported by the available schema.”
This helps RUBE_SEARCH_TOOLS return more relevant tool candidates and reduces unnecessary follow-up questions.
Common failure modes to watch
The main failure modes are:
- Skipping tool discovery and assuming schemas.
- Trying to execute before the
composio_searchconnection is active. - Providing vague goals that return broad or irrelevant tool matches.
- Ignoring missing required fields from the discovered schema.
When reviewing the agent’s plan, check that it names the discovered tool slug, required fields, connection status, and execution order before it acts.
Iterate after the first output
After the first discovery result, refine the prompt with the actual schema names returned by Rube. For example: “Use the discovered tool slug X. Fill required field Y with this value. Leave optional field Z blank unless it improves result relevance. Return only the top 10 results.”
This turns the composio-search-automation guide from a generic discovery workflow into a controlled execution loop.
Repository-level improvements worth adding
If you maintain or fork the skill, useful improvements would include a short example transcript, a minimal “discovery only” prompt, a “confirm before execute” prompt, and troubleshooting notes for inactive connections. The current skill is clear but compact; adding examples would reduce setup friction without changing its core workflow.