DevGlobe MCP server for consent-aware developer discovery

[sajeetharan]

Problem

A website can help developers find one another, but it does not let AI agents participate in that discovery. To make "developers and AI agents connect" literal, external agents need a standard, permission-aware interface to query DevGlobe without scraping profiles.

Proposed solution

Publish a DevGlobe MCP server that lets compatible agents discover public developer profiles and explain why a result matches a user's request.

Initial MCP tools

• search_developers: search public profiles by skills, location, agent stack, and collaboration availability
• get_developer_profile: retrieve a public, privacy-filtered profile by login
• get_rank_context: retrieve transparent global/country/language rank context
• prepare_introduction: validate and preview an introduction request; sending is handled by a separate consent workflow

Acceptance criteria

• Implement a versioned MCP server with documented tool schemas and structured responses
• Allow anonymous access only to explicitly public fields with conservative rate limits
• Support authenticated access using revocable credentials from Feature: User authentication with API keys #49
• Apply field-level privacy from Feature: Privacy controls for developers #47 consistently across web, API, and MCP responses
• Return source timestamps and explainable match reasons
• Prevent bulk profile export and add abuse/rate-limit controls
• Treat profile text as untrusted data and prevent it from changing tool instructions
• Publish setup examples for Claude Code, Codex, VS Code, and other MCP clients
• Add contract, authorization, privacy, rate-limit, and prompt-injection tests

Product safeguards

• The MCP server must never expose email addresses, private activity, raw embeddings, API keys, or hidden agent-stack entries.
• Read access does not grant permission to contact a developer.
• Tool responses must distinguish observed, verified, and self-declared claims.

Dependencies and related issues

• Depends on AI Agent Stack on developer profiles #72 and Agent-aware developer discovery and filters #73
• Authentication: Feature: User authentication with API keys #49
• Privacy controls: Feature: Privacy controls for developers #47
• Public profiles: Feature: Public shareable developer profiles #50