report

Generate the HTML report

Three forms:

• no args → report for the active repo
• <slug> → report for a specific repo (need not be active)
• --all → one report per repo this plugin has ever initialised

The script writes to ${repo_dir}/reports/<UTC-timestamp>.html and emits the absolute path on stdout. Reports are self-contained (inline CSS, no external assets), print-friendly, and safe to share — finding-content fields are HTML-escaped.

bash "${CLAUDE_PLUGIN_ROOT}/scripts/build-report.sh" "$@"

After the script returns:

1. Print the output path so the user can open it in a browser (xdg-open, open, or scp / serve from nginx).
2. If the user asked for a PDF / DOCX / PPTX format and the host has a converter installed, mention the follow-up command they can run:
   • PDF (any host with weasyprint): weasyprint <html-path> <pdf-path>
   • PDF (chromium-based): chromium --headless --no-sandbox --print-to-pdf=<pdf-path> file://<html-path>
   • DOCX (with pandoc): pandoc -f html -t docx <html-path> -o <docx-path>
   • PPTX (with pandoc): pandoc -f html -t pptx <html-path> -o <pptx-path>

Future releases of this plugin will wire the converter calls directly into the skill once the dependency choice is settled. For now the HTML is the canonical artefact.

Notes

• The report only includes findings that exist in the data directory; it does not re-scan or re-triage. Run /auto-audit:tick until the queue is drained before generating a final report if you want the latest state.
• Reports are append-only — every invocation writes a fresh timestamped file. Old reports are kept; ls ${repo_dir}/reports/ shows the history.
• The activity log section preserves the full iterations stream (one line per state transition). If a finding bounced between confirmed ↔ pr_rejected ↔ confirmed (a fixer retry), that history is visible.