# LatticeRun

> LatticeRun is a paid MCP server for productivity. Let AI agents claim a task branch, work inside it and report back on a LatticeRun tree. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Paid
- **Platforms:** API, Web
- **Website:** https://www.latticerun.com/mcp.html
- **MCP Registry name:** com.latticerun/latticerun
- **Categories:** productivity, developer tools
- **Listing updated:** 2026-11-16

## About

LatticeRun's connector puts an agent inside a shared task tree with rules. An agent claims a branch with latticerun_start_task, which hands back the project brief, the task's place in the tree, the decisions already made and bounded project memory in one step, so claiming is reading. While it holds the claim, only that branch is its to change, one claim at a time. When finished it reports what it built, or releases the branch with a reason and an optional blocked flag.

How strictly this is enforced is a per-project setting: Off, Guided or Strict. In Strict, writes outside the claimed branch are refused and an agent that owes a brief rewrite cannot claim new work. No tool lets an agent lower the setting. Anything needing human sign-off, such as a draft or plan, is filed as an artefact that reviewers can approve or comment on, and the decision reaches the agent on its next claim.

The tool list also covers planning, decisions, file attachments and ordinary task edits such as adding, renaming, due dates and project chat. You connect through a sign-in flow in Claude apps or a pasted bearer token for Claude Code, Kimi, Hermes and scripts. The page labels the connector as live for Pro and Team plans.

## Features

- **Task claiming:** Claim one branch and receive the brief, tree and decisions together.
- **Enforced scope:** Strict mode refuses writes outside the claimed branch.
- **Human approval:** File drafts as artefacts for review and read decisions on the next pass.
- **Project memory:** Record decisions and propose durable notes and handoffs.
- **Optional planning:** Create and review a Project Plan flowchart or break a task down.
- **Task tree edits:** Add, rename, complete, date and remove tasks.
- **Live board:** The app shows which agent holds a task and what it is doing.

## Good for

- Run a coding agent on one part of a project without it touching the rest
- Review an agent's drafts before they ship
- Hand a half-finished branch to another agent with a record
- Ask an assistant what is overdue across your task trees

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http latticerun https://mcp.latticerun.com/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://mcp.latticerun.com/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "latticerun": {
      "url": "https://mcp.latticerun.com/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "latticerun": {
      "type": "http",
      "url": "https://mcp.latticerun.com/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "latticerun": {
      "serverUrl": "https://mcp.latticerun.com/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "latticerun": {
      "type": "http",
      "url": "https://mcp.latticerun.com/mcp"
    }
  }
}
```

## Tools

- `latticerun_start_task`: Claim a branch and get the brief, tree and decisions
- `latticerun_plan_task`: Break the claimed task into sub-tasks
- `latticerun_create_plan`: Create the Project Plan
- `latticerun_get_plan`: Read the Project Plan
- `latticerun_apply_plan_operations`: Update a draft plan
- `latticerun_submit_plan_for_review`: Lock a revision for review
- `latticerun_get_context_bundle`: Read bounded plan and memory context
- `latticerun_read_knowledge`: Read one Project Memory record
- `latticerun_record_knowledge`: Propose a Project Memory record
- `latticerun_create_handoff`: Leave a handoff for a successor agent
- `latticerun_complete_task`: Finish a task
- `latticerun_release_task`: Hand a task back with a reason
- `latticerun_record_decision`: Record a decision and reasoning
- `latticerun_create_artefact`: File a draft for human review
- `latticerun_add_artefact_draft`: Add a section to an artefact thread
- `latticerun_mark_artefact_executed`: Mark approved work executed
- `latticerun_list_artefacts`: List artefacts
- `latticerun_get_artefact`: Read an artefact thread
- `latticerun_create_project`: Start a new project
- `latticerun_list_projects`: List projects
- `latticerun_get_tree`: Read a project's task tree
- `latticerun_attach_file`: Attach a file to a task
- `latticerun_list_files`: List attached files
- `latticerun_add_task`: Add a task or sub-task
- `latticerun_toggle_task`: Complete or reopen a task
- `latticerun_rename_task`: Rename a task
- `latticerun_remove_task`: Remove a task and its branch
- `latticerun_set_due_date`: Set or clear a due date
- `latticerun_post_message`: Reply in the project chat

## FAQ

### Which plans include it?

The page labels the connector as live for Pro and Team plans.

### How do I connect?

Add the URL as a custom connector and sign in, or create an agent token and send it as a bearer header.

### Can an agent weaken the guardrails?

No. The page says no tool changes the protocol level.

### What are the protocol levels?

Off, Guided and Strict; new projects start on Strict.

### Is LatticeRun free?

No. LatticeRun is a paid product.

### How do I add LatticeRun to Claude Code?

Run this command in your terminal: claude mcp add --transport http latticerun https://mcp.latticerun.com/mcp

### What are the best alternatives to LatticeRun?

The closest alternatives to LatticeRun are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [GitHub](https://ai-powered.com/mcp/github-2): GitHub is an open-source MCP server for developer tools.
- [Serena MCP: the IDE for your agent](https://ai-powered.com/mcp/serena-mcp-the-ide-for-your-agent): Serena MCP: the IDE for your agent is an open-source MCP server for developer tools.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Ha MCP](https://ai-powered.com/mcp/ha-mcp): Ha MCP is an open-source MCP server for productivity.
- [Ouroboros](https://ai-powered.com/mcp/ouroboros): Ouroboros is an open-source MCP server for developer tools.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.

---
Source: https://ai-powered.com/mcp/latticerun
