adrapid-automation
by ComposioHQadrapid-automation is a Claude skill for Adrapid workflow automation through Composio Rube MCP. It guides agents to install Rube MCP, verify an active Adrapid connection, search live tool schemas first, and execute with less guesswork.
Score: 64/100. This is an acceptable but limited listing candidate: directory users can understand that it enables Adrapid automation through Composio/Rube MCP and how to begin tool discovery and connection setup, but they should expect to rely heavily on live tool search and external toolkit schemas because the repository provides little Adrapid-specific workflow detail.
- Valid skill frontmatter declares the triggerable purpose and MCP requirement: automating Adrapid tasks via Rube MCP with `mcp: [rube]`.
- Prerequisites and setup steps are clear enough for an agent to check `RUBE_SEARCH_TOOLS`, manage the Adrapid connection, and confirm ACTIVE status before acting.
- The skill explicitly instructs agents to search tools first for current schemas, reducing the risk of stale tool calls.
- No support files, examples, scripts, or references beyond the single SKILL.md and external toolkit docs, so users get little evidence of tested Adrapid-specific workflows.
- The guidance is mostly a generic Rube MCP discovery/connection pattern and does not document concrete Adrapid tasks, parameters, edge cases, or expected outputs.
Overview of adrapid-automation skill
What adrapid-automation does
adrapid-automation is a Claude skill for running Adrapid operations through Composio’s Rube MCP server. Its main value is not a fixed workflow script; it teaches the agent to discover the current Adrapid tool schemas first, verify the Adrapid connection, and then execute the right Rube MCP tools for the user’s requested task.
Best fit for Workflow Automation users
The adrapid-automation skill is best for users who already use Claude with MCP and want an agent-assisted way to operate Adrapid without manually checking Composio toolkit schemas each time. It fits workflow automation tasks where tool availability, required fields, and execution plans may change, because the skill explicitly requires RUBE_SEARCH_TOOLS before action.
What makes this skill different
Many automation prompts assume tool names and parameters are stable. This skill is designed around discovery-first execution: search tools, inspect returned schemas, check connection status, then call the appropriate Adrapid tool. That makes it safer for a Composio/Rube environment where integrations can expose different tool slugs or input requirements over time.
Key adoption requirement
Do not install adrapid-automation unless your client can use the Rube MCP server. The skill requires RUBE_SEARCH_TOOLS and an active Adrapid connection through RUBE_MANAGE_CONNECTIONS. If your environment cannot add https://rube.app/mcp as an MCP server, the skill will read like guidance but cannot execute real Adrapid actions.
How to Use adrapid-automation skill
adrapid-automation install and setup context
Install the skill from the Composio skills repository with your skill manager, for example:
npx skills add ComposioHQ/awesome-claude-skills --skill adrapid-automation
Then configure Rube MCP in your AI client by adding:
https://rube.app/mcp
After MCP is available, verify that RUBE_SEARCH_TOOLS responds. Use RUBE_MANAGE_CONNECTIONS with toolkit adrapid to check whether the Adrapid connection is ACTIVE. If it is not active, follow the returned authentication link before asking the agent to run any Adrapid workflow.
Inputs the skill needs from you
For strong adrapid-automation usage, give the agent the business goal, the Adrapid object or operation you expect, relevant IDs or names, and any guardrails. Weak input is: “Update my Adrapid campaign.” Stronger input is:
“Use adrapid-automation to find the current Adrapid tools, check my connection, and update campaign Q4-Retargeting only if it is active. Do not create new campaigns. Show the discovered tool schema before execution and ask if a required field is missing.”
This improves output because the skill can map your goal to live Rube tool schemas instead of guessing tool names or required fields.
Recommended workflow for first run
Start by reading composio-skills/adrapid-automation/SKILL.md; it is the core source file and there are no extra scripts, references, or README files in the skill folder. Then run the workflow in this order:
- Ask the agent to call
RUBE_SEARCH_TOOLSfor your specific Adrapid use case. - Review returned tool slugs, schemas, and pitfalls.
- Ask the agent to call
RUBE_MANAGE_CONNECTIONSforadrapid. - Confirm the connection is active.
- Execute only the selected tool with the schema returned in the same session.
Prompt pattern that works well
Use a prompt that forces discovery and confirmation:
“Use the adrapid-automation skill for [specific Adrapid task]. First call RUBE_SEARCH_TOOLS with this use case: [task]. Then check RUBE_MANAGE_CONNECTIONS for toolkit adrapid. If connected, summarize the tool you plan to call, required inputs, and any risks before executing. If the schema does not contain enough information, ask me for the missing fields.”
This pattern keeps the agent aligned with the skill’s main constraint: current schemas must come from Rube, not memory.
adrapid-automation skill FAQ
Is adrapid-automation only for Composio users?
Yes. The skill is specifically written for Composio’s Rube MCP and the Adrapid toolkit. It is not a standalone Adrapid SDK, browser automation package, or API wrapper. Its execution path depends on Rube tools such as RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS.
Can I use it without an active Adrapid connection?
You can install and read the skill, but real automation requires an active Adrapid connection. The setup flow expects the agent to call RUBE_MANAGE_CONNECTIONS with toolkit adrapid and follow the returned authentication flow if the connection is not active.
How is this better than a normal prompt?
A normal prompt may invent Adrapid actions or rely on outdated assumptions. The adrapid-automation skill tells the agent to search live tool schemas first, then execute based on the returned tool definitions. That is the main reason to use the adrapid-automation skill instead of a generic “automate Adrapid” instruction.
When should I not use this skill?
Do not use it for non-Adrapid workflows, offline planning where no MCP tools are available, or high-risk production changes without a human review step. The source skill is concise and does not include detailed rollback procedures, test fixtures, or safety policies, so add your own approval gates for destructive or expensive actions.
How to Improve adrapid-automation skill
Improve adrapid-automation results with better task framing
The most important improvement is specificity. Include the Adrapid entity, desired operation, selection criteria, constraints, and whether the agent may execute or should only draft a plan. For example, say “pause ads matching X after showing the candidate list” rather than “clean up ads.” This gives RUBE_SEARCH_TOOLS a sharper use case and reduces irrelevant tool matches.
Common failure modes to watch
The biggest failure mode is skipping tool discovery and assuming a tool schema. The second is trying to execute before the Adrapid connection is active. A third is giving vague goals that require the agent to infer business rules. Prevent these by explicitly requiring discovery, connection check, schema summary, and confirmation before execution.
Iterate after the first output
After the first tool search, ask the agent to compare returned tools and explain which one best matches the task. If required fields are missing, provide them in a structured form: campaign name, ID, date range, target status, allowed changes, and stop conditions. Iteration should refine the exact tool call, not restart from a generic prompt.
Repository improvements worth adding
The upstream skill would be more adoption-ready with example prompts for common Adrapid operations, a short troubleshooting section for inactive connections, and safer execution patterns for update or delete actions. A small set of “plan first, execute after approval” examples would make adrapid-automation easier for workflow automation teams to use in production.