29 Releases 37 Tags

RSS Feed

• v1.6.0 f75f165911

  Compare

  v1.6.0 Stable

  tomas.kracmar released this 2026-04-22 07:55:21 +00:00 | 33 commits to main since this release

  v1.6.0 — Valkey caching + async queue for LLM scaling

  New features

  • Valkey caching for LLM responses (/api/ask and /api/events/{id}/explain)
    • Ask cache TTL: 1 hour
    • Explain cache TTL: 24 hours
  • Async queue (arq) for LLM requests to prevent timeouts at scale
    • POST /api/ask?async_mode=true returns a job_id immediately
    • GET /api/jobs/{job_id} polls for results
  • Valkey replaces Redis as the in-memory store (BSD-licensed fork)

  Infrastructure

  • New worker service in Docker Compose (runs arq jobs.WorkerSettings)
  • New redis/valkey service with healthchecks
  • REDIS_URL configuration setting

  API changes

  • AskRequest added async_mode: bool field
  • AskResponse added job_id: str | None field

  Dependencies

  • Added redis and arq

  Downloads

  • Source Code (ZIP)
  • Source Code (TAR.GZ)