I’m using several AI coding harnesses across different repos, and the thing that kept breaking down was not code editing — it was review artifacts scattered everywhere (Markdown reports in chat, HTML mockups, JSON manifests, etc.).
So I built a small OSS dashboard to give agents a shared local presentation surface: any harness can write a report/manifest, and the dashboard renders it as a report with Markdown, diffs, tables, comparisons, approvals, etc.
Repo: https://github.com/TaylorFinklea/harness-deck
Feedback on the structure/manifests welcome.
Top comments (0)