Imported from rhtmdgks/xtb-lab-harness (AGENTS.md). Install upstream with npx skills add rhtmdgks/xtb-lab-harness. Copyright stays with the author.

xTB Lab Harness

MCP tool server for xTB computational chemistry. Not a web app.

Scope

In scope: xTB CLI wrapper, output parser, MCP tools (calculate_candidate, calculate_candidate_batch, generate_evidence_table), JSON schemas, evidence logs under data/runs/.
Out of scope (MVP): Next.js, FastAPI, DB, auth, Gemini API inside the server. MAS orchestration lives on the MCP client side.

Layout

src/xtb_lab_harness/   Python package
examples/              Sample .xyz inputs
data/runs/             Per-candidate xTB run logs (gitignored contents)
docs/                  Architecture, tool contracts, report-format.md

Conventions

Return structured Pydantic models / dicts from tools; preserve raw_evidence paths.
xTB is invoked via subprocess in xtb/runner.py; parsing in xtb/parser.py.
MCP transport: stdio (python -m xtb_lab_harness.server or project script).

# after Install: the listing is in your workspace's routing pool - a plan picks it for its slot curl -s -X POST https://api.opensmartroute.ai/api/v1/route -H 'Authorization: Bearer $OSR_API_KEY' -H 'Content-Type: application/json' -d '{"text": "...", "plan": true}'

Install xtb-lab-harness and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

xtb-lab-harness

About

xTB Lab Harness

Scope

Layout

Conventions

Use it

Install xtb-lab-harness and it becomes one more candidate the router can pick - when it fits.

Manifest

Reviews