Skip to content

Releases: REPPL/Testimony

v0.4.0

Choose a tag to compare

@github-actions github-actions released this 24 Jul 16:38
249dc68

What's Changed

  • Harden the pipeline against untrusted sessions (v0.4.0) by @REPPL in #23

Full Changelog: v0.3.0...v0.4.0

v0.3.0

Choose a tag to compare

@github-actions github-actions released this 22 Jul 09:04
d965f4c

What's Changed

  • Add CHANGELOG covering v0.1.0 and v0.2.0 by @REPPL in #14
  • Fix correctness bugs: audio-only merge, capture write errors, verdict grouping by @REPPL in #15
  • CI: run on merge_group (prerequisite for the merge queue) by @REPPL in #16
  • Fix session dir collision and whitespace-line handling by @REPPL in #17
  • Fix validation upper bound and missing-t0 merge (burst 4) by @REPPL in #19
  • Propagate bursts 3 and 4 to main by @REPPL in #20
  • Fix 13 bugs from a multi-round review; record four intents by @REPPL in #21
  • Robustness fixes to convergence; v0.3.0 CHANGELOG by @REPPL in #22

Full Changelog: v0.2.0...v0.3.0

v0.2.0

Choose a tag to compare

@github-actions github-actions released this 18 Jul 13:49
9a7a56d

What's Changed

  • Add checksum-verified one-line installer by @REPPL in #4
  • transcribe: default WhisperX VAD to silero by @REPPL in #5
  • Restructure the record: brief, intents, end-user docs (re-target to main) by @REPPL in #7
  • Implement testimony record: one-command capture by @REPPL in #8
  • Implement testimony analyze + review: first-pass analysis, human-verified by @REPPL in #9
  • Security hardening: network, injection, symlink, supply-chain by @REPPL in #10
  • Security: sanitise finding id and verdict fields in report/review by @REPPL in #11
  • Security: verdict-echo sanitisation, BiDi controls, evidence cap by @REPPL in #12
  • CI: abcd-managed gates, tag-triggered release, attested installer by @REPPL in #13

Full Changelog: v0.1.0...v0.2.0

v0.1.0 — walking skeleton + transcribe

Choose a tag to compare

@REPPL REPPL released this 17 Jul 16:40
a452c76

First public release: the testimony CLI with the capture → transcribe → merge → report pipeline.

  • demo — instrumented settings app; streams normalised interactions (data-testid selectors) + raw rrweb archive into a fresh session directory
  • transcribe — local ASR (WhisperX preferred, whisper.cpp fallback) turns a voice recording into a word-aligned transcript.jsonl; raw audio never leaves the machine
  • merge / report — one session-relative timeline, rendered as an aligned Markdown record
  • record — stub (managed capture lands later)

Static binaries (no runtime dependencies) for macOS and Linux, arm64 + amd64. Verify with SHA256SUMS. Transcription requires local tools: ffmpeg plus WhisperX or whisper.cpp — the installer offers to set these up, including admin-free local installs.

Assisted-by: Claude:claude-fable-5