Overture MCP Server
Interactive visual flowcharts for AI agent planning before code execution.
Data last scanned yesterday · Reviewed 3 weeks ago
Overview
Overture serves as an intermediary layer that intercepts the planning phase of various AI coding agents. It transforms internal agent logic into visual flowcharts, allowing developers to review and approve proposed changes in a structured format prior to implementation. By visualizing dependencies, branching logic, and required context, the tool aims to reduce wasted tokens and unnecessary file modifications caused by misunderstood prompts. Users interact with these plans through an interface that prioritizes transparency before the agent writes or executes any code.
Our verdict
Overture receives a security score of 29/100, indicating a high level of risk. This rating is driven by the lack of a read-only mode, as the tool currently operates with full write access, and the fact that it is a community-maintained project rather than an official vendor release. Users should exercise caution and review the project's codebase before granting it access to sensitive environments.
- Reduces token waste by requiring plan approval before code execution.
- Provides visual transparency for dependencies and task branching.
- Features zero direct dependencies for a reduced supply-chain footprint.
- Includes active, detailed documentation within the repository.
- Lacks a read-only mode, granting the server full write access.
- Maintained by a single contributor, posing potential bus-factor risks.
- Missing formal security review or official vendor maintenance.
- Authentication methods have not yet undergone independent verification.
Setup
{
"mcpServers": {
"overture": {
"args": [
"-y",
"overture"
],
"command": "npx"
}
}
}Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.
- No read-only mode — running this grants full read/write access.
- No built-in authentication — anyone with access to this config can invoke every tool it exposes.
Tools
| Tool | Description | Risk |
|---|---|---|
| submit_plan | Submit complete plan as XML | medium |
| get_approval | Wait for user approval (blocks until approved) | low |
| update_node_status | Update node status + output during execution | medium |
| plan_completed | Mark plan as successfully completed | medium |
| plan_failed | Mark plan as failed with error message | medium |
| check_rerun | Check if user requested a node re-run | low |
| check_pause | Check if user paused execution | low |
| get_resume_info | Get full context for resuming a paused plan | low |
| request_plan_update | Request incremental plan modifications | medium |
| create_new_plan | Signal creation of a new plan | medium |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| ChatGPT | ||||
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf |
Frequently asked questions
›Which AI coding agents are compatible with Overture?
Overture works with any MCP-compatible AI coding agent, including Claude Code, Cursor, Cline, GitHub Copilot, and Sixth AI.
›How do I access the Overture visual planning interface?
Once installed and integrated with your agent, Overture automatically opens at http://localhost:3031 to display your interactive project plan.
›Does Overture start writing code immediately upon receiving a prompt?
No, Overture intercepts the agent's planning phase to present a visual flowchart, and the agent will not write any code until you approve the plan.
›How do I configure GitHub Copilot to use Overture?
You must create an .mcp.json file in your project's .vscode folder using the 'servers' configuration key, as GitHub Copilot requires VS Code 1.99+.
Alternatives
Changelog
Badge
Maintain this server? Add the live badge to your README.