MCP Server

ProdE provides an MCP server that connects your coding agent to your codebase understanding and your project context. When you set it up, you choose one of two modes, depending on whether you also want project management in your editor.

Choose a Mode

Mode	What your agent gets	Choose it when
Knowledge + Projects (default)	Codebase understanding and read/write project access: requirements, workstreams, documents, plan artefacts (flows, contracts, tasks), decisions, and pipeline status	You use ProdE Projects and want your agent to build with full spec and codebase context, and to keep the plan current. This is the right choice for most people.
Knowledge only	Codebase understanding alone: codebase Q&A, cross-repo understanding, code and symbol search, and documentation lookup	You only need codebase understanding and do not use ProdE Projects.

The Knowledge + Projects mode is a superset: it includes every Knowledge tool plus the project tools. Start there unless you specifically want to leave Projects out.

Endpoints

Each mode has its own endpoint. You configure exactly one.

Knowledge + Projects: https://api.prode.ai/core/api/v1/mcp/projects/mcp/
Knowledge only: https://api.prode.ai/code-parsing/v1/mcp/

Setting Up

Open the MCP Server section of your ProdE account and pick your mode and editor. ProdE generates a ready-to-paste configuration for Cursor, GitHub Copilot (VS Code), Windsurf, or Claude Code, with your authentication token and the correct format for that editor already included. Copy it into your editor's MCP configuration and you are connected.

The integration uses secure, token-based authentication. Communication is encrypted and your data remains private.

Example Configuration

A sample configuration for Claude Code (~/.claude.json), using the default Knowledge + Projects mode:

{
  "mcpServers": {
    "prode": {
      "type": "streamable-http",
      "url": "https://api.prode.ai/core/api/v1/mcp/projects/mcp/",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>"
      }
    }
  }
}

To use Knowledge only instead, keep everything the same and swap the url to https://api.prode.ai/code-parsing/v1/mcp/.

Available Tools

The tables below list every tool the server exposes. The Knowledge tools are available in both modes. The Projects tools are added in the default Knowledge + Projects mode.

Knowledge Tools

Powered by your Knowledge Layer: the indexed, AI-enriched representation of your repositories. Instead of generic code suggestions, your agent can answer codebase questions, trace how services relate across repositories, and search code precisely.

Tool	Description
get_all_repositories	List all repositories available in your Knowledge Layer
get_high_level_context	Get AI-summarized documentation and code symbol discovery across repos
get_documentations	Fetch full auto-generated documentation content
search_file_contents	Regex-based search across repository files
read_file	Read specific files from your repositories
get_folder_structure	View repository directory structure
find_symbol_usages	Find all usages of a function, class, or variable

Projects Tools

Added in Knowledge + Projects mode. These let your agent both read project context and write back to it while you work, so it builds against the real spec and keeps the plan current without leaving your editor. The full project tree is editable: requirements (including acceptance criteria), workstreams, documents, plan artefacts (flows, contracts, tasks), and decisions.

Read & Explore

Tool	Description
list_projects	List all projects accessible to you
get_project_overview	Get a comprehensive overview of a project
list_workstreams	List the workstreams in a project
list_requirements	List requirements with optional filters (workstream, status, priority, dev status)
list_documents	List project documents, optionally filtered by tags
list_artifacts	List the artefacts in a project
list_blockers	List blockers with optional filters
read_entity	Read any project entity by its ID: a requirement, workstream, document, plan artefact, or blocker
read_all_decisions	Read every decision logged on a project
get_pipeline_stats	Get pipeline and progress statistics
get_engineering_phases	Get the status of the engineering planning phases
ask_project_question	Ask a natural-language question about a project

Create

Tool	Description
create_project	Create a new project
create_workstream	Create a workstream
create_requirement	Create a requirement under a workstream
create_document	Create a project document
create_artifact	Create an artefact: an HTML widget, a flow, a contract, or a task
create_decision	Log a decision

Update & Edit

Tool	Description
update_requirement	Update a requirement's fields (status, dev status, design status, priority) and its acceptance criteria
update_workstream	Update a workstream's header fields
update_document	Update a document's title, tags, or summary
update_artifact	Update an artefact's name, description, or metadata
update_decision	Update or resolve a decision
edit_content	Make surgical edits to the body of a requirement, workstream, document, or plan artefact (flow, contract, task)
delete_entity	Delete a project entity

Status & Pipeline

Tool	Description
set_stage_status	Mark a pipeline stage (PRD, Product Spec, UI/UX, Engineering, Development, QA) complete or incomplete
set_engineering_phase_status	Set the status of an engineering planning phase

MCP Server

On this page