Skip to content

CHANGELOG.md

Latest commit

 

History

History
147 lines (105 loc) · 7.71 KB

CHANGELOG.md

File metadata and controls

147 lines (105 loc) · 7.71 KB

Changelog

0.8.16 — 2026-03-17

  • Adopted the shared bin/lint script and wired make lint / make check to the March repo convention for Rust repos.
  • Refreshed the README, homepage, CLI reference, and helper docs so the public release docs match the current BioMCP command surface.
  • Expanded skills/SKILL.md for article graph helpers, enrich, batch, chart discovery, chart flags, and updated drug interaction guidance.
  • Updated the Kuva chart blog post with checked-in SVG outputs for all nine worked examples plus placeholder references for terminal screenshots.

0.8.15 — 2026-03-11

  • Fixed the planning-docs CI path regression so release validation uses the repo-local planning fixtures by default instead of an Ian-local absolute path. This is the fix from PR #191 that unblocks release packaging on GitHub Actions.
  • Refreshed the public discovery docs so search all is taught as the unified cross-entity entry point in the README and docs index. This is the docs alignment from PR #190.
  • Breaking MCP runtime change: renamed the MCP execution tool from shell to biomcp; update MCP clients and demos to call biomcp after tools/list.

0.8.14 — 2026-03-10

  • Promoted remote Streamable HTTP in newcomer docs with a dedicated getting-started page for biomcp serve-http.
  • Added a runnable PEP 723 demo client at demo/streamable_http_client.py for /mcp, including a tools/list flow and biomcp version tool call.
  • Surfaced the canonical remote HTTP routes more prominently across public docs: /mcp, /health, /readyz, and /.
  • Kept serve-sse migration guidance visible while shipping the Streamable HTTP release/docs/demo verification package together.

0.8.13 — 2026-03-09

  • study survival now reports Kaplan-Meier median survival, 1/3/5-year survival landmarks, and two-group log-rank p-values.
  • Fixed survival median calculation: replaced raw follow-up median with Kaplan-Meier median survival.
  • study compare --type expression now reports Mann-Whitney U and p-value.
  • Added study download <study_id> and study download --list for local cBioPortal datahub installs.
  • Hardened study download: no total body timeout for large archives, path-like study ID rejection, stream-to-disk download, and cleanup on failed extraction.
  • Refreshed skills/SKILL.md with real study command usage, including cohort, filter, and survival --endpoint examples.

0.8.12 — 2026-03-07

  • Added study subcommand: local cBioPortal study adapters with CLI query surface (download, cohort, survival, compare).
  • Added study filter: multi-omics cross-table sample joins across mutation, CNA, expression, and clinical data.
  • Added Fisher exact p-values to co-occurrence pair analysis.
  • Added skill installer auto-discovery: scans existing agent config directories in priority order and supports the .agents/skills/ cross-tool standard.
  • Refreshed skill guidance and reference material: adopted the measured SKILL.md refresh and added drift validation for examples, schemas, and jq snippets.
  • CLI quality pass: positional search arguments, multi-token input handling, variant fallback, NCT ID validation, bare-entity shortcuts, staged trial fill, article request timeout, and --until alias.

0.8.11 — 2026-03-06

  • Added expression section to gene output (GTEx tissue-specific TPM data).
  • Added druggability section to gene output (DGIdb drug-gene interactions and categories).
  • Added clingen section to gene output (gene-disease validity and dosage sensitivity).
  • Added evidence URLs (_meta.evidence_urls) to all entity output — includes Ensembl, OMIM, NCBI Gene, and UniProt links where available.
  • Added spec/ BDD documentation suite with 54 passing executable specifications.
  • Unified article search: fan-out across PubTator3 and Europe PMC in parallel with PMID deduplication, source-grouped rendering, and --source <all|pubtator|europepmc>.
  • Added infinite cache mode via BIOMCP_CACHE_MODE=infinite for offline/demo workflows.
  • Consolidated CI into a single job; streamlined release pipeline.

0.8.10 — 2026-03-04

  • Polished search all output: GWAS trait relevance filtering, clinical significance sorting, uninformative variant suppression.
  • Gene-anchored article search via GENE_PROTEIN field for higher precision.
  • Added trial search improvements (--mutation boolean handling and --criteria support).
  • Added PyPI trusted publisher setup and release workflow updates.

0.8.9 — 2026-03-03

  • Added search all cross-entity command — parallel fan-out across major entities (genes, variants, diseases, drugs, trials, articles, pathways, PGx, GWAS, and adverse events) with counts-first display and HATEOAS deep-dive links.

0.8.8 — 2026-03-02

  • Output cleanup across entity renderers for consistent markdown formatting.
  • Improved help discoverability and completeness across all subcommands.
  • Rewrote footer output with streamlined next-step suggestions.

0.8.7 — 2026-02-27

  • Improved help text discoverability and completeness for all CLI subcommands.

0.8.6 — 2026-02-27

  • Fixed trial filter accuracy: age-based post-filtering, exclusion-criteria detection, and pathway source correctness.

0.8.5 — 2026-02-26

  • Fixed install.sh to resolve latest tag to a release that has downloadable assets.
  • Hardened input safety with SSRF prevention on user-supplied URLs.
  • Enforced pagination and validation contracts across search endpoints.

0.8.4

  • Simplified CI/CD workflow triggers to avoid redundant post-merge runs.
  • Added runner-aware actions/cache@v4 protoc caching in CI, docs deploy, and release workflows.
  • Pinned workflow protoc installation to 28.3 with cache-aware conditional setup.

0.8.3

  • Added inclusion-first eligibility post-filtering to reduce exclusion-only false positives in trial matching.
  • Applied cryptography dependency security updates and refreshed release artifacts.
  • Tightened CLI/docs consistency around trial filter behavior and output contracts.

0.8.2

  • Added location-level facility and geo post-filtering for more accurate trial site matching.
  • Embedded git tag version metadata in the binary for reliable biomcp version output.
  • Improved pagination and result-window handling across search command families.

0.8.1

  • Expanded clinical-grade trial filtering with ~45 new search flags and validation rules.
  • Added MCP tool auto-generation to reduce drift between CLI capabilities and server tool surface.
  • Upgraded help/discoverability content and list-based guidance for agent-driven workflows.

0.8.0

Complete rewrite from Python to Rust. Single static binary, no runtime dependencies.

Highlights

  • Single-binary CLI and MCP server — no Python, no pip, no virtual environments
  • 15 biomedical data sources, unified command grammar, compact markdown output
  • 14 embedded skills (guided investigation workflows)
  • MCP server (stdio + SSE) with tool and resource support
  • HTTP proxy (serve-http) for multi-worker shared rate limiting
  • Production installer with SHA256 verification (5 platforms)
  • Progressive disclosure: search returns summaries, get returns full detail with selectable sections
  • NCBI API key support for improved rate limits
  • 429 retry with Retry-After backoff

Data sources

MyGene.info, MyVariant.info (ClinVar, gnomAD, CIViC, OncoKB), ClinicalTrials.gov, NCI CTS API, PubMed/PubTator3, MyChem.info, Monarch/MONDO, Reactome, UniProt, OpenFDA FAERS, PharmGKB/CPIC, GWAS Catalog, Monarch (phenotypes), AlphaGenome (gRPC).

Breaking changes from Python BioMCP

  • Python package biomcp-python is no longer maintained
  • MCP tool names and signatures have changed
  • Configuration via environment variables only

0.7.3

Legacy Python BioMCP. See branch python-0.7.3 for source code.