Skip to content

Scriveno 3.2.4

Choose a tag to compare

View all tags
@hannsxpeter hannsxpeter released this 08 Jun 19:18
· 14 commits to main since this release
v3.2.4
cd59526

What changed

  • Proof is now replayable. Golden workflow fixtures replay the flagship novel, technical runbook, and publishing-package paths, including expected next commands.
  • Voice DNA has a regression harness. A paired-sample eval now checks expected voice markers and forbidden generic patterns without pretending to be an objective prose score.
  • Detector anxiety stays inside the authenticity workflow. Outside AI-detector reports are context for /scr:voice-check, /scr:originality-check, /scr:line-edit, and /scr:polish, not targets for rewriting or humanizer-style laundering.
  • Host parity has capture targets. Claude Code, Codex, and one standard command runtime now have explicit capture-ready directories and queue metadata. They are not marked verified until real host transcripts are recorded.
  • Installer maintainability improved. Runtime/profile metadata moved into a small shared module, and installed commands now get a generated fallback for the Next commands: contract.
  • The first journey is tighter. The recommended starter loop is now first-run, demo, next, draft, review, and save before the wider command surface opens up.
  • Proof claims are badged. Major trust claims now point to evidence levels such as shipped, install-tested, artifact-proven, replay-tested, host-capture-ready, host-captured, or external-tool-dependent.

Why it matters

Scriveno keeps its versatility, but the claims behind that versatility are easier to inspect and harder to regress. This release turns several previously readable proof artifacts into executable regression evidence while staying honest about host-runtime work that still needs real captures.

Assets 3
Loading