> For the complete documentation index, see [llms.txt](https://docs.apolo.us/index/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.apolo.us/index/apolo-mcp/readme.md).

# Apolo MCP documentation

Apolo MCP is a local [Model Context Protocol](https://modelcontextprotocol.io/) server that gives an AI client typed, bounded access to Apolo platform operations. It uses the configured Apolo identity and permissions: normally a local `apolo login`, or an isolated job's `APOLO_PASSED_CONFIG`. It does not shell out to the CLI, return credential values to the model, or grant additional permissions. It can create service accounts, but their one-time tokens are written directly to a protected file or Apolo secret instead of entering the model conversation.

The server is intended for discovering platform context, running and observing workloads, operating Applications, and managing supported project resources. It does not provide arbitrary shell or HTTP access, Kubernetes administration, identity or RBAC administration, model-visible secrets, interactive terminal streams, or a shared remote service.

## Components

Apolo MCP ships two complementary capabilities:

1. **Local MCP server.** A local stdio process exposes typed, bounded Apolo operations to an MCP client. The server owns platform access, explicit context resolution, output bounds, credential protection, mutation policy, and lifecycle journaling. See the [capability matrix](/index/apolo-mcp/capabilities.md) and generated [MCP tool reference](/index/apolo-mcp/capabilities/tools.md).
2. **Agent skills catalogue.** Packaged skills for Codex and Claude Code teach the client how to combine MCP tools into task-oriented workflows such as context discovery, research jobs, Flow workloads, Applications, resource management, and isolated R\&D sessions. See the generated [skills catalogue](/index/apolo-mcp/capabilities/skills.md).

The MCP server is the controlled execution layer. The skills catalogue is the workflow layer that guides an agent in using that execution layer safely and consistently.

## Start here

1. [Install and configure Apolo MCP](/index/apolo-mcp/getting-started/installation.md) and check its prerequisites.
2. Read the generated [safety model](/index/apolo-mcp/getting-started/safety.md) before enabling writes.
3. Explore the [capability matrix](/index/apolo-mcp/capabilities.md) to understand what is supported.
4. Use the generated [tool reference](/index/apolo-mcp/capabilities/tools.md) and [skills catalog](/index/apolo-mcp/capabilities/skills.md) for exact interfaces and workflows.
5. Follow a guide for [platform context](/index/apolo-mcp/guides/platform-context.md), [workloads](/index/apolo-mcp/guides/workflows.md), [Applications](/index/apolo-mcp/guides/applications.md), or [isolated full-mode operation](/index/apolo-mcp/guides/full-mode-service-account.md).

The repository [README](https://github.com/neuro-inc/apolo-mcp/tree/master/README.md) contains the short project summary. This documentation is the maintained usage and capability contract for the current release.