scientific-toolkit: one MCP server fronting nine repos that refuses to fabricate results
Emergency-Tutor-7 · reddit · 2026-11-22
A Reddit user open-sourced scientific-toolkit (GitHub: Furox-Art/scientific-toolkit, PyPI: scientific-toolkit-mcp 0.4.1): a single MCP server with nine adapters exposing independently versioned research repos (model engine, literature evidence, verification gates, scaffolding, etc.). The key design: every adapter reports what actually happened — a missing dependency returns NOTINSTALLED on the call instead of a plausible fabricated answer, addressing the failure mode that makes MCP servers untrustworthy. The author notes it's a hub, not a merge: nine adapters mean nine failure modes, and if you need just one tool, install that repo directly.
More from coding & agent
- Walmart and Muse to launch first agent-to-agent agentic commerce integration over MCP — AccBalanced · 2026-11-22
- Zero-degradation Strata tuning: fixing layer_split lifts decode ~10-15% and halves CPU expert misses — PathfinderTactician · 2026-11-22
- Cross-provider adversarial review: hedge shared model blind spots — craigbalding · 2026-11-22
- Inference Stack Launches One-Command Self-Hosted LLM Serving with vLLM, LiteLLM and llm-d — AccBalanced · 2026-11-22
- Users complain GPT-5.5 writes bloatware and duplicates functions — yacineMTB · 2026-11-22
- SWE-bench cost analysis: tiny Inkling 2 Mini costs 3-5x more per task than DeepSeek V4.1 Flash — hyudryu · 2026-11-22