Menu
Expert guide for testing bioinformatics MCP servers. Covers pytest, DRY_RUN modes, and PatientOne simulation scenarios.
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Audits all Markdown files in the precision-medicine-mcp repo against four canonical principles: (A) server Python code is the authoritative tool count, (B) server-registry.md is the authoritative server/tool summary, (C) all three patient outcomes (PAT001/PAT002/PAT003) must be represented consistently, (D) no content block may appear in more than one MD file (DRY). Use this skill whenever a new version of the platform ships, a new server is added, a new patient use case is validated, or before any major doc update. Trigger phrases: "audit the docs", "check docs are correct", "doc cleanup", "canonical check", "are the docs up to date", "DRY check on markdown".
Ensures all development meets clinical compliance and security standards. Focuses on HIPAA-friendly logging, audit trails, risk assessment, and scientific nomenclature.
Guide for managing patient data and bioinformatics datasets (h5ad, CSV). Standardizes data loading, preprocessing, and path management.
Expert guide for project-specific documentation and clinical reporting. Maintains consistency across persona-based guides and automated reports.
Expert guide for GCP deployment, security, and hospital-specific infrastructure. Covers Cloud Run, IAM, VPC, and secrets management for bioinformatics.
Expert guide for developing the Precision Medicine Streamlit dashboard and LLM providers. Covers provider abstraction, trace visualization, and UI components.
| name | precision-bio-tests |
| description | Expert guide for testing bioinformatics MCP servers. Covers pytest, DRY_RUN modes, and PatientOne simulation scenarios. |
This skill provides guidelines for validating the complex biological and clinical logic in this repository.
pytest)Tests live in two locations:
tests/unit/mcp-[server]/ (e.g., tests/unit/mcp-spatialtools/)servers/mcp-[name]/tests/ (e.g., servers/mcp-perturbation/tests/)Both patterns are in active use. Check both locations when looking for existing coverage.
.fn Testing PatternFastMCP Pattern: Never call @mcp.tool() decorated functions directly. Always use the .fn attribute to bypass the MCP protocol layer:
result = await server.perturbation_predict.fn(gene="BRCA1", cell_type="T-cell")
This pattern is used across 9 test files in the project. Always follow it when writing new tests.
Always test both modes:
DRY_RUN=True (default) — returns synthetic/simulated data, no real bio tools neededDRY_RUN=False — exercises real logic (monkeypatch external dependencies in CI)Some test suites use fixture files (e.g., tests/unit/mcp-multiomics/fixtures/ contains sample CSV files). Large datasets should be mocked or referenced from data/patient-data/.
The gold standard for integration testing is the PatientOne workflow (docs/reference/testing/patient-one/):
For browser-based or chat-based testing:
docs/reference/testing/quick-test-prompts.md.fn pattern used for all tool function calls in testsUse this skill when: