Skip to content

v0.3.16

Latest
Latest

Choose a tag to compare

View all tags
@ForisKuang ForisKuang released this 08 Oct 18:08
v0.3.16
fc96de2
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

What's Changed

Gene query API and API keys (for scripts)

  • Personal ACGC API keys for scripts, sent as Authorization: Bearer acgc_…. Only a hash of each key is stored. Each key has its own rate limit (60/min by default) and is re-checked against the email-domain allowlist on every request. A key can't create, list or revoke keys. (#108)
  • In-app API keys page at /api-keys to create (shown once), list and revoke keys. Admins see everyone's keys. (#113)
  • Gene query endpoints that return only the classification and rationale, plus a view_url link to the full report: POST /v1/genes/query (up to 50 genes), GET /v1/genes/{symbol}, and background jobs at /v1/genes/query/jobs. (#107)
  • Gene query fixes: 404 only when HGNC or Ensembl confirms a gene doesn't exist, and 503 when the lookup itself fails. /v1/annotate/gene no longer returns a link to a run that wasn't saved. (#114)
  • New API user guide in docs/api.md, linked from the README. (#115)

Login

  • Run links (/?run=<id>) now survive login: users with an expired session sign in and return to the run. The redirect target is restricted to paths on this site. (#106)

OpenEvidence

  • The OpenEvidence card no longer times out on slow first lookups. A cache miss answers "pending" right away while the lookup runs in the background, and the card checks back until it's ready. Paid calls are deduplicated. (#102)
  • The default OpenEvidence model is now osler. (#105)
  • Fixed widget metadata leaking into OpenEvidence card text. (#104)

Literature

  • PubMed preprints are labeled end to end. They rank below peer-reviewed research, prompts treat them as weaker evidence, and evidence cards show a "Preprint – not peer-reviewed" badge. No papers are dropped. (#112)

Benchmarks and CI

  • osler vs darwin OpenEvidence benchmark (#103), a fix for its tests (#109), and accurate time-to-first-token measurement (#111).

Deploy notes

  • No migration: the api_keys table is created at startup.
  • New settings all have defaults; see .env.example (API_KEY_*, OPENEVIDENCE_SIDECAR_*).
  • Keycloak login stays off until KEYCLOAK_CLIENT_ID and KEYCLOAK_CLIENT_SECRET are set; Google login is unchanged.
Assets 2
Loading