Skip to content

TownReporter 0.6.29 — scheduled discovery and evidence review

Choose a tag to compare

@scottconverse scottconverse released this 08 Sep 01:28
af6e78e

Changes

  • Owners can configure a disabled-by-default daily scan of up to 12 accepted sources using one selected local model or existing Claude/Codex subscription. Reservations and result writes are fenced against duplicate work, stale settings and lost permission. It files leads only; it does not draft or publish.
  • Draft jobs now commit the draft, lead status and notes, audit event and job completion atomically. Lost claims and withdrawn permission stop stale final writes, and terminal bookkeeping cannot turn completed work into a failure.
  • Multi-source drafts preserve complete structured provenance, findings and unanswered questions through saving and publication instead of truncating JSON.
  • The story workbench can review recorded findings against private captured passages and save explicit editor judgments. Stale draft or evidence views refuse saves; public evidence routes remain scoped to the public newsroom and exact published references.
  • Draft reporting keeps the originating newsroom's paper identity, captures and provenance. Mismatched and foreign capture/version references are excluded.

Install

Download TownReporter-0.6.29-windows-x64.zip, extract it, and open Install TownReporter.cmd. See the Windows installation guide. Windows x64 is the packaged platform; macOS/Linux source setup remains documented. The ZIP contains source and the installer, which downloads its private runtime dependencies during installation.

Source: af6e78e9862f82b3ba84b8c7b5520ba101da128c.
ZIP: 6,969,637 bytes (6.97 MB).
SHA-256: ec6dadda0149bb021401488735f44636cb5e5faeae4d6005e003a454c9366ea2.

Verification and scope

The reviewed source passed all application CI jobs, both Windows package/install checks, and Pages publication. The candidate archive matched all 763 tracked blobs at the exact source commit; its manifest and checksum sidecar matched the archive.

The evidence-review audit disclosed degraded reviewer independence and retained two Minor findings and one Nit as nonblocking watchlist items. This release does not claim exhaustive claim coverage, independent fact-checking, automatic publication, measured 60-minute daily operation, a universal sub-hour installation, or Halo deployment.