keap-automation

Overview of keap-automation skill

What keap-automation does

keap-automation is a Claude skill for running Keap CRM operations through Composio’s Rube MCP server. Its main value is not a fixed Keap script; it teaches the agent to discover the current Keap tool schemas first, verify the connection, then execute CRM actions with the right tool names and required fields.

Use it when you want an agent to help with Keap contact, CRM, marketing, or sales-operation tasks without guessing API parameters from memory.

Best fit for CRM Operations teams

The keap-automation skill is best for CRM Operations, RevOps, automation builders, and support teams that already use Keap and want AI-assisted workflows through MCP. It is especially useful when your task depends on live tool availability, because the skill explicitly requires RUBE_SEARCH_TOOLS before execution.

Good-fit jobs include finding the correct Keap operation, checking the active Keap connection, preparing structured inputs, and running a workflow only after the schema is known.

What makes it different from a generic prompt

A generic prompt may ask Claude to “update Keap contacts” and rely on assumed field names. keap-automation pushes the agent into a safer sequence:

• discover tools with RUBE_SEARCH_TOOLS
• manage or verify the Keap connection with RUBE_MANAGE_CONNECTIONS
• use returned schemas rather than invented arguments
• keep session context across discovery and execution

That discovery-first pattern is the main differentiator and the main adoption reason.

Adoption considerations before install

This skill requires Rube MCP and an active Keap connection. It is not useful as an offline Keap reference, and it does not include support scripts, examples, or extra reference files beyond SKILL.md. If your client cannot connect to MCP tools, or your Keap auth cannot be completed, the skill will not execute real operations.

How to Use keap-automation skill

keap-automation install context

Install the skill from the Composio skill collection, then configure Rube MCP in your AI client. The repository path is composio-skills/keap-automation, and the key file to read first is SKILL.md.

A typical install command is:

npx skills add ComposioHQ/awesome-claude-skills --skill keap-automation

Then add https://rube.app/mcp as an MCP server in your client configuration. The upstream skill states that no API key is needed for the MCP endpoint itself, but Keap still must be connected through Rube.

Required setup before first workflow

Before asking for a Keap action, confirm these conditions:

1. RUBE_SEARCH_TOOLS is available.
2. RUBE_MANAGE_CONNECTIONS can check toolkit keap.
3. The Keap connection status is ACTIVE.
4. The agent has searched for the specific use case before calling an execution tool.

The skill’s important operational rule is: always call RUBE_SEARCH_TOOLS first. This matters because Composio tool slugs, required fields, and recommended execution plans can change.

Turn a rough goal into a strong prompt

Weak prompt:

“Update my Keap contacts.”

Better prompt for keap-automation usage:

“Use the keap-automation skill. First call RUBE_SEARCH_TOOLS for the use case ‘find or update Keap contact by email and add a tag’. Check the Keap connection with RUBE_MANAGE_CONNECTIONS. If active, show me the required schema before executing. Contact email: [email protected]. Desired change: add tag Webinar Lead. Do not create a duplicate contact unless the discovered tool explicitly supports safe upsert behavior.”

This stronger prompt improves output because it gives the agent the business goal, lookup key, intended mutation, duplicate-handling constraint, and permission boundary.

Practical workflow to follow

Start each session with tool discovery using your exact task, not a broad phrase. For example, search for “create Keap contact with phone and company,” “list Keap contacts by tag,” or “update opportunity stage” rather than only “Keap operations.”

After discovery, ask the agent to summarize:

• selected tool slug
• required inputs
• optional inputs worth setting
• risks or missing fields
• whether the connection is active
• proposed execution step

Only approve execution once the schema and target records are clear. This is especially important for bulk updates, tagging, campaign enrollment, and contact creation, where a small field mistake can affect many CRM records.

keap-automation skill FAQ

What does keap-automation need to run?

It needs an MCP-capable client, Rube MCP configured at https://rube.app/mcp, access to RUBE_SEARCH_TOOLS, and an active Keap connection managed through RUBE_MANAGE_CONNECTIONS with toolkit keap.

Without those pieces, the skill can explain the workflow but cannot perform live Keap operations.

Is keap-automation beginner-friendly?

It is beginner-friendly for users who understand their Keap goal, but it is not a one-click CRM automation product. The user should be able to describe the target record, desired change, and whether the action is read-only or write-enabled. Beginners should ask the agent to show the discovered schema and execution plan before approving any changes.

When should I not use this skill?

Do not use keap-automation when you need a static Keap tutorial, a custom integration outside Rube MCP, or a guaranteed list of Keap API fields without live discovery. Also avoid using it for large destructive updates unless you can provide filters, record IDs, rollback expectations, and a review step.

How is it different from using Keap directly?

Keap’s UI is better for manual review and visual workflows. The keap-automation skill is better when you want an agent to translate an operational goal into the correct Composio/Rube tool call, especially across repetitive CRM tasks. It reduces schema guesswork, but it does not replace human approval for sensitive CRM changes.

How to Improve keap-automation skill

Make keap-automation prompts more specific

The fastest way to improve results is to provide exact operational context. Include the object type, lookup method, fields to change, allowed side effects, and whether the agent should execute or only prepare.

Stronger input pattern:

• “Object: contact”
• “Lookup: email”
• “Action: add tag”
• “If not found: report only”
• “Before execution: show schema and selected tool”
• “After execution: summarize changed record IDs”

This gives the skill enough information to choose safer tools and avoid ambiguous CRM actions.

Common failure modes to prevent

The main failure mode is skipping discovery and assuming tool arguments. Prevent this by explicitly saying: “Search tools first and use the returned schema.”

Other common issues include inactive Keap auth, vague record matching, accidental duplicate creation, and approving bulk changes without previewing filters. For write operations, ask the agent to confirm the target count before execution whenever the discovered tools allow it.

Iterate after the first output

After the first plan, do not simply approve if anything is unclear. Ask follow-up questions such as:

• “Which required fields are still missing?”
• “Can this tool update by email, or does it require contact ID?”
• “Will this create a contact if no match is found?”
• “What is the safest read-only check before the write step?”

These questions improve the execution path and reduce CRM data-quality risk.

Read the repository with the right expectations

For a keap-automation guide, read SKILL.md first and focus on the sections named Prerequisites, Setup, Tool Discovery, and Core Workflow Pattern. There are no extra scripts or reference folders in the current skill package, so most of the practical behavior comes from following the documented Rube MCP sequence carefully rather than from local code.