> ## Documentation Index
> Fetch the complete documentation index at: https://clearskies.cc/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# What your AI can access over MCP

> The categories of Context Graph data and Clearskies capabilities available to any AI you connect over MCP: CRM records, activity, schema, research, and building tools.

When you connect an AI to Clearskies over [MCP](/docs/building-with-clearskies/mcp-server), it gets a set of tools for reading your [Context Graph](/docs/unifying-gtm-data/customer-context-graph) and building in Clearskies. This page describes what those tools let a connected AI do, whether that's Claude, ChatGPT, or a platform like n8n or Retool. For connection steps, see [MCP Server](/docs/building-with-clearskies/mcp-server).

Each category below names a few representative tools rather than listing every one, the definitive list is always what your connected AI sees at runtime.

## CRM records

Read the accounts, contacts, and deals in your graph, and answer questions across them.

* **List and read records**: `accounts_list`, `contacts_list`, and `deals_list` return your core CRM records, and `crm_records_list` covers any other object type in your graph
* **Drill into an account**: `account_get_contacts` and `account_get_deals` pull one account's contacts or deals directly
* **Aggregate**: `records_aggregate` runs counts and other aggregations across records, pipeline by stage, deals by owner, without pulling every record into the conversation

## Schema discovery

Before querying, a connected AI can look up which object types and fields exist in your workspace, so it queries real fields instead of guessing.

* **Object types**: `object_definitions_list` returns the object types in your graph
* **Fields**: `object_get_fields_schema` returns the field schema for one object type
* **Search**: `schema_search` finds relevant objects and fields without pulling the full schema

## Activity and communications

The calls, emails, and meetings connected to your accounts, not just that they happened, but what was said.

* **List and search**: `events_list` and `events_search` find calls, emails, and meetings across the graph or for a specific account
* **Read contents**: `events_get_contents` returns the substance of an event, a call transcript, an email body
* **Look ahead**: `calendar_get_upcoming` shows upcoming meetings, useful for call prep

## Team and workspace

* **People**: `employees_list` returns the people on your team, so an AI can attribute activity to the right owner
* **Identity**: `identity_get` resolves who the connected user is
* **Slack**: `channels_list` returns the Slack channels available in your workspace

## Support tickets

`support_tickets_list` returns support tickets connected to your graph, so account health questions can include what's happening in support.

## Engineering activity

`github_activities_list` returns GitHub activity connected to your graph, putting engineering work alongside CRM and communication data.

## Deep research

For questions that span many records and sources, a connected AI can hand the work to Clearskies instead of assembling it call by call: `deep_research` starts a long-running research job across the graph, and `deep_research_status` polls it until the result is ready.

## Building workflows and agents

Beyond reading data, a connected AI can take Clearskies workflows and agents through their full lifecycle, create, validate, test, and publish, using the `workflows_*` and `agents_*` tools, with `workflow_capabilities_get` and `agent_capabilities_get` telling it what your workspace supports before it builds. See [Manage workflows from Claude or ChatGPT](/docs/workflows/manage-via-mcp) and [Manage agents from Claude or ChatGPT](/docs/agents/manage-via-mcp) for how that works.

To put these tools to work from an assistant, see [Build in Claude or ChatGPT](/docs/building-with-clearskies/build-in-claude-or-chatgpt).