MCP Overview

Edsger implements the Model Context Protocol (MCP) to enable seamless integration with AI agents and tools like Claude. The MCP server exposes a comprehensive API for managing the full software development lifecycle.

What is MCP?

The Model Context Protocol is a standardized way for AI assistants to interact with external systems. It provides:

• Resources: Data that AI can read (products with full feature details)
• Tools: Actions that AI can perform (CRUD operations across all entities)
• Prompts: Pre-defined conversation templates (system prompts, workflow templates)

Available Resources

The MCP server exposes product data as resources:

Use resources/list to discover available products and resources/read to fetch full product data.

Available Tools

The MCP server advertises tools that AI agents can call:

Feature Management

User Stories

Test Cases

Checklists

Agents & Workflows

Full API Reference

Beyond the advertised tools, the MCP server supports a comprehensive set of JSON-RPC methods:

Products & Features

• products/list, products/get
• features/list, features/get, features/update, features/claim, features/info

User Stories & Test Cases

• user_stories/list, user_stories/create, user_stories/delete, user_stories/update_status, user_stories/existing
• test_cases/list, test_cases/create, test_cases/delete, test_cases/update, test_cases/update_status, test_cases/existing

Test Reports

• test_reports/create, test_reports/get, test_reports/list, test_reports/latest, test_reports/update
• test_report_results/create

Checklists

• checklists/list, checklists/create, checklists/update, checklists/delete, checklists/phase, checklists/validate
• checklist_items/create, checklist_items/update, checklist_items/delete
• checklist_results/create, checklist_results/list
• checklist_item_results/create

Feedbacks

• feedbacks/get, feedbacks/list_for_feature, feedbacks/list_for_product
• feedbacks/create, feedbacks/update, feedbacks/delete, feedbacks/resolve

Approvals

• approvals/requires_approval, approvals/get_assignees, approvals/feature_approvals, approvals/create

GitHub Integration

• github/developer_config, github/installation_token, github/config_and_token, github/config_and_token_by_product

Branches & Pull Requests

• branches/list, branches/get, branches/create, branches/update, branches/delete, branches/current, branches/clear
• pull_requests/list, pull_requests/get, pull_requests/create, pull_requests/update, pull_requests/delete, pull_requests/clear

Feature Audit Logs

• feature_audit_logs/create, feature_audit_logs/list

Chat

• chat/channels/get_or_create, chat/channels/list
• chat/members/add, chat/members/remove, chat/members/list
• chat/messages/list, chat/messages/send, chat/messages/send_ai, chat/messages/send_system
• chat/messages/pending, chat/messages/claim, chat/messages/mark_processed
• chat/read_status/mark_read, chat/read_status/unread_count

Tasks

• tasks/list_for_product, tasks/create, tasks/update, tasks/delete, tasks/product_members

CLI Sessions & Logs

• cli_sessions/register, cli_sessions/heartbeat, cli_sessions/deregister, cli_sessions/list
• cli_logs/batch, cli_logs/list

Notifications

• notifications/send_phase_completion_email

Growth & App Store

• growth/campaigns, growth/save_analysis, growth/save_video_metadata
• app_store/configs/list, app_store/configs/get, app_store/configs/upsert, app_store/configs/save_listings, app_store/configs/save_screenshots, app_store/configs/update_status

Getting Started

1. Create an MCP token
2. Configure your AI agent with the token
3. Connect to the Edsger MCP endpoint

MCP Endpoint

https://your-project.supabase.co/functions/v1/mcp

Example Integration

Claude Desktop / Claude Code

{
  "mcpServers": {
    "edsger": {
      "url": "https://your-project.supabase.co/functions/v1/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_MCP_TOKEN"
      }
    }
  }
}

Authentication

All requests require a bearer token in the Authorization header. Tokens are validated via the validate_mcp_token RPC function, which checks:

• Token validity and expiration
• Allowed resources based on token configuration
• Rate limits (per-minute and per-hour)

All requests are logged to mcp_access_logs for monitoring and auditing.