review-report-writer¶
Turn an aggregated pre-push review result into a self-contained, human-readable HTML report — verdict banner, per-dimension cards, findings grouped by severity with file:line locations, and the scope/scanner/fleet metadata. It renders findings it is GIVEN; it never re-reviews code, never re-runs a scanner, and never invents a finding. Writes one .html file outside the repo tree and returns its path.
At a glance¶
| Model | inherits the session's |
| Tools | Read Write Bash |
| Author | navjyotnishant |
| Source | agents/review-report-writer.md |
Writes files
This agent produces a file itself, rather than returning content for the skill to write — one of only two that do.
When it runs¶
The pre-push-review umbrella has aggregated five dimension verdicts and wants a report a human can open in a browser.
write the review report
The pre-push-review skill spawns this agent last, after aggregation, with the structured findings; the agent returns the file:// path the skill prints.
What it returns¶
Write the file, then return only:
- The absolute path, and a
file://URL for it. - One line confirming what you rendered: the aggregate verdict, the count of findings by severity, and the number of dimensions.
- Anything you had to mask, omit for missing data, or could not render.
Do not restate the findings in your reply — they are in the file, and the caller already has them. Your value is the artifact, not a second summary of it.
Spawned by¶
Derived, not declared
No agent file records which skills call it — this list is recovered from the skill definitions at build time, so it cannot go stale.
See the /pre-push-review pipeline for where this fits in the whole run.