highlevel-automation
by ComposioHQhighlevel-automation is a Claude skill for automating HighLevel workflows through Composio Rube MCP with schema-first tool discovery, connection checks, and safer execution guidance.
This skill scores 68/100, which means it is acceptable for listing but should be presented as a lightweight MCP workflow guide rather than a complete automation package. Directory users get enough evidence to understand when to install it—Highlevel automation through Rube MCP—and enough operational guidance to start safely, but they should expect to discover most task-specific details dynamically rather than from bundled examples or reference files.
- Clearly identifies the intended trigger: automating Highlevel operations through Composio's Highlevel toolkit via Rube MCP.
- Provides actionable prerequisites and setup steps, including adding the Rube MCP endpoint and confirming an ACTIVE Highlevel connection.
- Strongly instructs agents to call RUBE_SEARCH_TOOLS first, which reduces schema guesswork and helps keep executions aligned with current tool definitions.
- No support files, scripts, references, or concrete worked examples are included beyond SKILL.md, so users must rely on Rube's returned schemas for task specifics.
- The skill depends on an active Rube MCP and Highlevel connection, and the excerpt shows possible naming inconsistency between RUBE_MANAGE_CONNECTIONS and RUBE_MANAGE_CONNECTION.
Overview of highlevel-automation skill
What highlevel-automation does
highlevel-automation is a Claude skill for automating HighLevel tasks through Composio’s Rube MCP server. Its main value is not a fixed list of actions; it teaches the agent to discover the current HighLevel tool schemas first, confirm the account connection, and then execute workflows through the available Rube tools.
Best fit for Workflow Automation users
Use the highlevel-automation skill when you want an AI agent to help with HighLevel operations such as contact, CRM, marketing, or account workflow tasks where the exact tool schema may change. It is best for users already working with Claude, MCP, Composio/Rube, and a connected HighLevel account.
Key differentiator: schema-first execution
The important pattern is “search tools before acting.” Instead of assuming outdated parameters, the skill directs the agent to call RUBE_SEARCH_TOOLS for the specific use case, read the returned tool slugs and input schemas, then proceed. This reduces failed calls caused by stale examples or guessed field names.
Adoption considerations
This is a lightweight skill: the repository path includes a single SKILL.md and no bundled scripts, references, or helper assets. That makes highlevel-automation easy to inspect, but it also means success depends on your MCP setup, active HighLevel authorization, and the quality of the task brief you provide.
How to Use highlevel-automation skill
highlevel-automation install and setup context
Install the skill from the Composio skills repository, then make sure your client can use Rube MCP:
npx skills add ComposioHQ/awesome-claude-skills --skill highlevel-automation
The skill requires the rube MCP server. Add https://rube.app/mcp as an MCP server in your client configuration. Then verify that RUBE_SEARCH_TOOLS is available. Before running HighLevel actions, use the connection management tool for the highlevel toolkit and complete the returned authorization flow if the connection is not active.
Inputs the skill needs from you
For reliable highlevel-automation usage, give the agent a clear business goal plus enough operational context to select the right HighLevel tools. Include the object type, action, filters, fields to update, safety limits, and whether the agent should preview before execution.
Weak prompt:
Update my HighLevel contacts.
Better prompt:
Use highlevel-automation for Workflow Automation. First discover current HighLevel tools with
RUBE_SEARCH_TOOLS. Then find contacts taggedWebinar-Leadcreated in the last 14 days, preview the count and sample records, and only after confirmation add them to the nurture workflow namedQ3 Demo Follow-up. Do not modify contacts without an email address.
Recommended workflow
Start by reading SKILL.md; it contains the complete operating pattern. In practice, a good run looks like this:
- Ask the agent to call
RUBE_SEARCH_TOOLSfor your exact HighLevel task. - Have it summarize available tool slugs, required fields, and pitfalls.
- Confirm the HighLevel connection is active through the Rube connection tool.
- Ask for an execution plan before write operations.
- Run read or preview steps first when the action affects many records.
- Execute only after schemas, filters, and target records are clear.
This sequence matters because the skill depends on live tool discovery, not a hardcoded HighLevel API map.
Practical prompt pattern
A strong prompt for the highlevel-automation skill usually has this shape:
Use the highlevel-automation skill. Discover current HighLevel tools first.
Goal: [business outcome]
Target records: [contacts/opportunities/campaigns/etc.]
Selection rules: [tags, dates, pipeline stage, location, status]
Action: [create/update/search/trigger workflow]
Required fields: [field names and values if known]
Safety: [preview first, max records, ask before writes]
Output: [summary, CSV-style table, errors, next steps]
Add naming details exactly as they appear in HighLevel when possible. If you do not know the field names, say so and ask the agent to infer only after reading the returned Rube schema.
highlevel-automation skill FAQ
Is highlevel-automation only for advanced users?
It is approachable if you already use Claude with MCP tools, but it is not a one-click automation package. Beginners can use it successfully by starting with read-only discovery tasks and asking the agent to explain each tool schema before any write action.
How is this better than an ordinary prompt?
An ordinary prompt may hallucinate HighLevel API fields or assume outdated tool names. highlevel-automation explicitly instructs the agent to use RUBE_SEARCH_TOOLS first, so the workflow is based on current Composio/Rube tool definitions returned at runtime.
When should I not use this skill?
Do not use it if you cannot connect Rube MCP, cannot authorize a HighLevel account, or need a standalone script that runs outside an AI/MCP client. It is also a poor fit for bulk destructive operations unless you can define strict filters, preview steps, and approval gates.
What repository files should I inspect first?
For this skill, inspect composio-skills/highlevel-automation/SKILL.md. There are no extra scripts or reference folders in the current structure, so the install decision should focus on whether the MCP requirements and schema-discovery workflow match your environment.
How to Improve highlevel-automation skill
Make highlevel-automation inputs more exact
The fastest way to improve results is to replace vague CRM language with operational constraints. Instead of “clean up leads,” specify the segment, the target field, acceptable values, excluded records, and what counts as success. This gives the agent enough information to choose the correct discovered tool and avoid broad unintended updates.
Avoid common failure modes
Most failures come from skipping discovery, acting before the HighLevel connection is active, or providing incomplete filters. For write operations, require the agent to state the selected tool slug, required schema fields, planned arguments, and expected effect before execution. This catches mismatched field names and accidental bulk changes early.
Iterate after the first output
Treat the first response as a planning pass. If the agent returns tool options, ask it to compare them and choose the safest one. If it finds records, ask for a sample and count before changing anything. If a tool call fails, paste the error back and ask the agent to re-check the schema through RUBE_SEARCH_TOOLS rather than guessing a fix.
Add local operating rules if your team needs them
Because highlevel-automation ships as a compact skill, teams may want to add their own usage conventions: maximum batch size, required approval for workflow enrollment, field naming notes, location/account IDs, or standard tags. These local rules can make repeated HighLevel automation safer and more consistent without changing the core schema-first pattern.