Skip to content

v0.6.5

Choose a tag to compare

@github-actions github-actions released this 13 Oct 09:00
· 116 commits to main since this release

[0.6.5] - 2026-10-13

Features

  • Improved embedding batching
  • (synthesize) Limit demo users' synthetic interviews per user
  • (auth) Add password reset via emailed magic link
  • (auth) Enforce preview features on the API
  • (db) Enforce foreign keys by default
  • (codebook) Carry merges in the codebook save
  • (embed) Add a live flag to stop embedding during interviews

Bug Fixes

  • (export) Infer the messages dataframe schema from every row
  • (migrations) Keep the testrun triggers through the started_by migration
  • (migrations) Let alembic own the touch-last_updated triggers
  • (participants) Return 404 for opt-out links that match nobody
  • (websockets) End live sessions before deleting their interviews
  • (interviews) Refuse stale test-run, participant and interview ids
  • (interviews) End the session when a write hits a deleted interview
  • (synthesize) Scope test-setup routes to the project in their path
  • (db) Stamp trigger-bumped last_updated with the app clock
  • (migrations) Register app_now on alembic's engine
  • (migrations) Stop the orphan sweep deleting children of live parents
  • (ws) Run AI interviews in the interview's own language

Performance

  • (db) Index every foreign-key column

Internal

  • Update just backup-db recipe to use sudo when running backup script on remote
  • (migrations) Merge the release/0.4.x hotfix migrations into main
  • (release) Let only the final-release rule set latest
  • (migrations) Name e5f2a91c4d80 as the cause of the lost testrun triggers
  • (users) Rename with_demo_features to with_preview_features
  • Keep main linear with rebase and fast-forward merges
  • (settings) Correct the database env-var prefix and Postgres notes
  • (migrations) Delete foreign-key orphans again
  • (settings) Regenerate config.schema.json
  • Record the per-process session registry and last_updated clocks
  • (just) Disable two-factor on fetched database copies

[0.6.4] - 2026-09-22

Internal

  • Update db justfile recipes to match new ENV structure
  • Fix service settings default values

[0.6.3] - 2026-09-22

Internal

  • Remove old dead translations module
  • Publish a staging tag on every release
  • Preview a feature branch on staging
  • Reenable pytest in release wf and run on pull requests
  • Better release version detection in wf
  • Bump lib version

[0.6.2] - 2026-09-17

Features

  • Implement better code exploration integration
  • Add similarity search based on coded turns

[0.6.1] - 2026-09-16

Features

  • Improve duration trimming for interviews by capping answer gaps to 5min

[0.6.0] - 2026-09-16

Features

  • Switch from old annotations to new codebook analytical framework

[0.5.8] - 2026-09-14

Features

  • Implement shared interview filtering for report, and add deduplicate_by_pid argument

Internal

  • Move interview filtering to standalone module

[0.5.7] - 2026-09-14

Internal

  • Bump lib version

[0.5.6] - 2026-09-14

Features

  • Add condition carrier counts to report

[0.5.5] - 2026-09-11

Internal

  • Move relabel resumed survey answers from scripts to alemibc migration

[0.5.4] - 2026-09-11

Internal

  • Bump lib version

[0.5.3] - 2026-09-11

Internal

  • (scripts) Relabel survey answers stored as free text

[0.5.2] - 2026-09-11

Features

  • Add groupby controls (interview/guide) and question filtering to all units

[0.5.1] - 2026-09-10

Features

  • Improve embeddings service
  • Add chunks for interview kinds

[0.5.0] - 2026-09-10

Features

  • Add report endpoint
  • Let comments be author specific and implement comment replies
  • Add comment authorship and permissions
  • (report) Add conditions to report payload
  • (monitoring) Change characters to log scale and duration from seconds -> minutes
  • Add embedding components
  • Implement new and better any-llm exception handling
  • Improved embedding chunk turn identification and grouping
  • Add umap embeddings and add language information to embedding chunks
  • Improve language filtering for embeddings
  • Implement pagination for embeddings endpoints
  • Add question filtering for data exploration
  • Add browsing with keywords
  • Improve keyword search match
  • Streamline transcript payloads
  • Add survey filtering for embeddings
  • Add order and filtering to browsing endpoints
  • Add sections as embedding kind

Bug Fixes

  • Cascade delete annotations when interviews are deleted
  • (build) Commit only release files in publish recipe
  • Update a users last_active status on token refresh
  • Better timeout handling for embedding client
  • Harden language type across api
  • Don't cast synthetic messages to ReceivedData before sumbission
  • Harden API authorization for annotations

Internal

  • Move histogram utility tools from monitor.py to shared histogram.py module
  • Bump ruff pre-commit
  • Update dev dependencies
  • Filter unnecessary umap warning
  • Refine db backup commands
  • Bump lib version