leadfeeder-automation
by ComposioHQleadfeeder-automation helps CRM Operations teams automate Leadfeeder workflows through Composio Rube MCP. Learn prerequisites, connection checks, tool discovery with RUBE_SEARCH_TOOLS, and safe usage patterns before running actions.
Score: 66/100. This is acceptable for listing because it gives agents a clear Leadfeeder + Rube MCP trigger, connection prerequisites, and a tool-discovery execution pattern that is more actionable than a generic prompt. For directory users, the listing should be treated as a lightweight integration wrapper rather than a complete Leadfeeder playbook: useful if they already use Rube/Composio, but thin on concrete Leadfeeder workflows and examples.
- Valid frontmatter declares the skill name, Leadfeeder automation purpose, and required `rube` MCP dependency, making the trigger and runtime dependency clear.
- Prerequisites and setup explain that Rube MCP must be connected, `RUBE_SEARCH_TOOLS` must be available, and an active Leadfeeder connection should be established through `RUBE_MANAGE_CONNECTIONS`.
- The skill repeatedly instructs agents to search tools first for current schemas, which reduces schema guesswork when using Composio's Leadfeeder toolkit.
- No support files, scripts, examples, or install command are provided beyond the single SKILL.md, so adoption depends on users already knowing how to configure MCP skills in their client.
- The guidance is mostly a generic Rube/Composio tool-discovery pattern rather than concrete Leadfeeder-specific automations, with limited practical task examples or edge-case handling.
Overview of leadfeeder-automation skill
What leadfeeder-automation does
leadfeeder-automation is a Claude skill for running Leadfeeder workflows through Composio’s Rube MCP server. It is designed to help an agent discover the current Leadfeeder tool schema, confirm the Leadfeeder connection, and execute CRM-adjacent tasks without guessing tool names or parameters.
The key behavior is not “call Leadfeeder directly.” The skill instructs the agent to use Rube MCP first, especially RUBE_SEARCH_TOOLS, so it can retrieve up-to-date Leadfeeder actions, required fields, execution plans, and pitfalls before attempting an operation.
Best fit for CRM Operations teams
The leadfeeder-automation skill is best for CRM Operations, RevOps, sales operations, and growth teams that already use Leadfeeder and want AI-assisted execution around account intelligence, company identification, lead routing, enrichment, or follow-up workflows.
It is most useful when your workflow depends on live Leadfeeder data and connected tooling rather than a static prompt. For example, a CRM Ops user might ask the agent to find available Leadfeeder actions, check whether the account is connected, then prepare or run a workflow that supports segmentation, lead qualification, or downstream CRM updates.
Important adoption requirements
This skill depends on Rube MCP. Your client must have https://rube.app/mcp configured as an MCP server, and the Rube tools must be available to the agent. The Leadfeeder toolkit connection must also be active through RUBE_MANAGE_CONNECTIONS.
The main differentiator is schema discovery. Because Composio tool schemas can change, the skill explicitly tells the agent to search available tools before execution. That makes it safer than hard-coding a Leadfeeder action name into a prompt, but it also means the skill is not useful in environments where MCP tools are disabled.
How to Use leadfeeder-automation skill
leadfeeder-automation install context
Install the skill from the GitHub skill repository if your environment supports Claude skills:
npx skills add ComposioHQ/awesome-claude-skills --skill leadfeeder-automation
Then configure Rube MCP in your AI client:
https://rube.app/mcp
Before expecting useful output, verify three things: RUBE_SEARCH_TOOLS is available, RUBE_MANAGE_CONNECTIONS can manage the leadfeeder toolkit, and the Leadfeeder connection status is ACTIVE. If the connection is not active, follow the auth link returned by Rube before asking the agent to run a business workflow.
Inputs the skill needs from you
A good leadfeeder-automation usage prompt should include the business outcome, relevant Leadfeeder objects or filters, the destination system if any, and whether the agent should only plan or actually execute.
Weak prompt:
“Use Leadfeeder to get leads.”
Stronger prompt:
“Use leadfeeder-automation to discover the current Leadfeeder tools through Rube MCP. Check whether the Leadfeeder connection is active. If active, find tools that can identify companies visiting our pricing and demo pages in the last 7 days. Return the available execution plan first; do not make changes until I approve.”
This improves output because it gives the agent a concrete use case for RUBE_SEARCH_TOOLS, defines a time window, clarifies intent, and prevents accidental execution.
Practical workflow for reliable execution
A strong leadfeeder-automation guide follows this sequence:
- Ask the agent to call
RUBE_SEARCH_TOOLSwith your specific Leadfeeder use case. - Have it inspect returned tool slugs, required input schemas, recommended plans, and warnings.
- Confirm the Leadfeeder connection with
RUBE_MANAGE_CONNECTIONS. - Ask for a short execution plan before running any operation that changes data or triggers downstream actions.
- Execute only after the agent has mapped your goal to the discovered schema.
This order matters. If the agent skips discovery, it may invent fields or use stale assumptions. If it skips connection checking, the workflow may fail halfway through with an authentication issue.
Repository files to read first
The upstream skill is compact and mainly contained in SKILL.md. Read it before install if you need to confirm the MCP requirement, setup sequence, and expected Rube tool calls. There are no extra resources/, rules/, references/, scripts/, or project README files in the skill folder, so the decision largely rests on whether the SKILL.md workflow matches your operating environment.
leadfeeder-automation skill FAQ
Is leadfeeder-automation useful without Rube MCP?
No. The skill requires Rube MCP and specifically depends on tools such as RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS. Without MCP access, it becomes ordinary guidance rather than an executable automation skill.
How is this better than a normal Leadfeeder prompt?
A normal prompt may ask the model to reason about Leadfeeder from memory. leadfeeder-automation pushes the agent to discover the current Composio Leadfeeder tool schema at runtime. That reduces hallucinated parameters and helps the agent adapt to the tools that are actually available in your connected environment.
Is it beginner-friendly?
It is beginner-friendly for users who can configure MCP and complete a Leadfeeder connection, but it is not a no-setup template. Non-technical CRM users may need help from an admin to add the Rube MCP server and authorize the Leadfeeder toolkit. After setup, prompts can be written in business language if they include clear workflow constraints.
When should CRM Operations not use it?
Do not use leadfeeder-automation for CRM Operations if your process requires fully audited, deterministic CRM changes without human review, or if your organization does not allow AI agents to access sales intelligence tools. Also avoid it for one-off strategic analysis where exported Leadfeeder data in a spreadsheet would be safer and simpler.
How to Improve leadfeeder-automation skill
Improve leadfeeder-automation prompts
The fastest way to improve leadfeeder-automation results is to make prompts operational. Include the target audience, filters, dates, page URLs, lead status, desired output format, and approval rules.
Example:
“Discover Leadfeeder tools for finding companies from the United States that visited /pricing or /demo at least twice in the past 14 days. Check connection status. Return the tool schema mapping, required fields, and a proposed execution plan. Do not update CRM records yet.”
This gives the agent enough context to search tools accurately and separate planning from execution.
Prevent common failure modes
The most common failure is skipping tool discovery and relying on assumed schemas. Make “search tools first” part of every serious request. The second failure is asking for broad automation without defining whether the agent may write, sync, tag, export, or only report. The third is missing authentication: if the Leadfeeder connection is not ACTIVE, fix that before debugging prompts.
For higher-risk workflows, require a dry run or plan-first response. Ask the agent to list the exact tool slug, required inputs, and expected side effects before execution.
Iterate after the first output
After the first plan or result, refine with business rules rather than rewriting the whole prompt. Add exclusions such as existing customers, competitors, low-fit countries, internal traffic, or companies already assigned to sales reps. If the output is too broad, add stronger thresholds such as visit count, page intent, company size, geography, or recency.
For CRM Operations, the best iteration loop is: discover tools, validate filters, preview results, approve execution, then document the final workflow so it can be reused consistently.