Kommo Automation

Overview of Kommo Automation skill

What Kommo Automation does

Kommo Automation is a CRM operations skill for controlling Kommo through natural-language requests backed by the Composio MCP toolkit. It helps an agent list pipelines and stages, filter leads, create or update deals, assign tasks, and work with custom fields without forcing the user to manually translate every action into API calls.

Best fit for CRM operators and revenue teams

This Kommo Automation skill is best for sales operations, founders, account managers, and support-to-sales teams that already use Kommo and want an assistant to handle repeatable CRM work. It is especially useful when your goal includes structured CRM actions: “find leads stuck in negotiation,” “create follow-up tasks,” or “update these deals with a custom field.”

What makes it different from a generic CRM prompt

A generic prompt can suggest CRM process ideas, but it cannot reliably act inside Kommo. This skill is built around Composio MCP access to Kommo tools such as KOMMO_LIST_LEADS_PIPELINES, KOMMO_LIST_PIPELINE_STAGES, and KOMMO_LIST_LEADS, so the assistant can request real CRM data and perform supported operations after authentication.

Important adoption requirements

You need a Kommo account and an AI client that can use the Composio MCP server. The source skill references the MCP endpoint https://rube.app/mcp and OAuth-based Kommo connection. The repository currently consists primarily of SKILL.md, so expect a compact skill definition rather than a large framework with scripts, examples, and test fixtures.

How to Use Kommo Automation skill

Kommo Automation install context

For a skill-directory install, use your client’s skill installation flow, for example:

npx skills add ComposioHQ/awesome-claude-skills --skill "Kommo Automation"

Then configure the Composio MCP server in the AI client:

https://rube.app/mcp

When prompted, connect your Kommo account through OAuth. The skill depends on the rube MCP integration, so installation alone is not enough; the assistant needs active MCP tool access before Kommo Automation usage will produce CRM actions instead of plain advice.

Inputs the skill needs to act correctly

Kommo requests work best when you provide identifiers or narrowing criteria. Useful inputs include pipeline name or pipeline_id, stage name or status, responsible user, lead date range, price range, target lead IDs, task due dates, and the exact custom field to read or update.

Weak prompt:

“Clean up my pipeline.”

Stronger prompt:

“In Kommo, list leads in pipeline 12 that have been in the negotiation stage for more than 14 days, group them by responsible user, and suggest follow-up tasks due tomorrow. Do not update records until I confirm.”

The stronger version gives the agent scope, filters, output format, and a safety boundary.

Suggested workflow for safe CRM operations

Start read-only, then move to updates. A practical Kommo Automation guide looks like this:

1. Ask the skill to list available pipelines and stages.
2. Confirm the correct pipeline_id and stage IDs.
3. List a small sample of leads using filters.
4. Review the assistant’s proposed changes.
5. Ask it to create tasks, update lead fields, or modify deals only after confirmation.

This pattern reduces accidental edits, especially in teams with multiple pipelines or similar stage names.

Repository files to read first

Read composio-skills/kommo-automation/SKILL.md first because it contains the available workflows, setup notes, tool names, and example prompts. There are no separate README.md, rules/, references/, or scripts/ files indicated in the repository preview, so the skill’s operating assumptions are concentrated in that one file. For tool-level detail, use the linked Composio Kommo toolkit documentation: https://composio.dev/toolkits/kommo.

Kommo Automation skill FAQ

Is Kommo Automation for CRM Operations or sales strategy?

Kommo Automation for CRM Operations is mainly for executing and organizing CRM work: listing leads, navigating stages, filtering records, creating tasks, and updating deal data. It can support sales strategy by surfacing pipeline information, but it is not a substitute for forecasting methodology, sales coaching, or custom analytics unless you provide the relevant CRM context and ask for analysis.

Can beginners use this skill?

Yes, if they understand basic Kommo concepts such as leads, pipelines, stages, tasks, and custom fields. Beginners should start with read-only prompts like “show my pipelines” and “list leads in this stage” before asking for updates. The main learning curve is not the skill syntax; it is knowing which Kommo records should be changed.

When should I not use Kommo Automation?

Do not use it for bulk destructive changes without a review step, ambiguous requests such as “fix all bad leads,” or workflows that require unsupported business logic outside Kommo. If your organization has strict CRM governance, add approval language to every prompt, such as “prepare the changes but do not execute them until I approve.”

How does it compare with using Kommo directly?

Kommo’s interface is better for visual browsing and one-off manual edits. The Kommo Automation skill is more useful when you want the assistant to combine search, filtering, summarization, and repeatable actions from a single instruction. It is strongest for operational batches where you can define clear criteria.

How to Improve Kommo Automation skill

Improve Kommo Automation results with precise scope

The biggest quality lever is scope. Instead of saying “update my leads,” specify the pipeline, stage, lead segment, dates, responsible owner, and desired outcome. If you do not know IDs, first ask: “List my Kommo pipelines and stages with IDs, then wait.” This gives the assistant the exact handles needed for later tool calls.

Add guardrails for risky updates

For create, update, or task-assignment requests, tell the assistant how to handle uncertainty. Strong guardrails include: “show a preview table first,” “limit to 20 records,” “skip leads missing an owner,” “do not overwrite non-empty custom fields,” and “ask before changing statuses.” These instructions materially reduce CRM cleanup mistakes.

Common failure modes to watch for

The most common issue is ambiguous naming: two pipelines may sound similar, or a stage name may not be unique. Another failure mode is asking for analysis that requires fields not exposed in the current result set. If an answer looks incomplete, ask the skill to retrieve the relevant custom fields or rerun the query with a narrower filter.

Iterate after the first output

Treat the first response as a discovery pass. A good iteration sequence is: list pipeline structure, filter leads, inspect a sample, define the update rule, preview affected records, then execute. This makes Kommo Automation more reliable than a single broad instruction and gives you a clear audit trail for CRM changes.