Deep Research MCP open-sources a pipeline to hook five research engines into Claude Code
PMinervini · x · 2026-10-03
Researcher PMinervini released deep-research-mcp, a Python MCP server that plugs multiple deep-research providers into Claude Code: OpenAI Deep Research (Responses API, gpt-6-sol, or experimental Codex OAuth with a ChatGPT subscription), Gemini Deep Research, Tavily Research, Allen AI's DR-Tulu, and the open-source Open Deep Research stack (smolagents-based).
His workflow: before committing to a research idea, let coding agents like Codex/Claude call these DR agents to check whether the idea already exists — "to make sure I'm not reinventing the wheel," while admitting recall is far from 100%. The thread also cites the classic joke: Gemini hails the idea as ICLR-spotlight-worthy, ChatGPT says it's nice, and Claude replies "Schmidhuber did this in 1993." The repo ships with CLI, agent skills, docs and tests.
More from coding & agent
- Lucy hits 94.7% Recall@10 on FinanceBench with Qdrant over 2.8M SEC/DART filings — qdrant_engine · 2026-10-03
- The AI slop loop: agents rewriting each other's junk, and thin context is the root cause — IgorCarron · 2026-10-03
- 8 core networking concepts every developer should understand — goyalshaliniuk · 2026-10-03
- Codex vs Claude Opus head-to-head: porting retro console games from CPU to GPU rendering — ssh4net · 2026-10-03
- Open-source iFixAi audits AI agents in 120s with 60 checks and an A-F grade — thisdudelikesAI · 2026-10-03
- Git doesn't fit parallel agents: worktrees eat disk, so agents will design the alternative — Al_Grigor · 2026-10-03