run-report

Generate a self-contained HTML closeout report summarizing shipped PRs, renders, gate numbers, and costs.

Updated Dec 27, 2025
One-click install
npx skills add https://github.com/HendrikGC02/Astroray --skill run-report-hendrikgc02
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: run-report
Source: https://github.com/HendrikGC02/Astroray/tree/main/.claude/skills/run-report
Command: npx skills add https://github.com/HendrikGC02/Astroray --skill run-report-hendrikgc02

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? After an orchestrator run or a significant package landing, the owner needs a single readable artifact that captures what shipped, what changed visually, and what the measured gate numbers were, without digging through PRs, ledgers, and test outputs manually. ## Core Features & Use Cases - Structured closeout report: Produces one self-contained HTML file with a fixed section order — headline, landed PRs, visual evidence, gate numbers, time and cost, and parked or failed items. - Visual evidence embedding: Embeds before/after render pairs as downscaled base64 data URIs with captions, sourced from test_results, verifier PNGs, and refbank renders. - Gate and cost tables: Records measured values versus thresholds verbatim in tables, plus wall time per phase, delegated model token cost, and CI minutes. - Use Case: At the end of an overnight Astroray orchestrator session, gather merged PRs via gh, pull gate numbers from spec Status lines, embed the changed renders, and deliver a styled HTML report the owner reviews alongside the standup. ## Quick Start Use the run-report skill to generate the closeout report for today's run, including landed PRs, before/after renders, gate numbers, and the time and cost breakdown.

Frequently Asked Questions about run-report

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate an HTML run report after an orchestrator session?▼

Invoke the run-report skill at the end of the run. It gathers merged PRs from gh, gate numbers from spec Status lines, renders from test_results, and timing from the ledger, then produces one self-contained HTML file delivered via SendUserFile.

What sections does the run report contain?▼

The report has six fixed sections in order: headline paragraph, landed PRs table, visual evidence with before/after renders, gate numbers versus thresholds, time and cost breakdown, and parked or failed items with tracking references.

How are render images embedded in the HTML report?▼

Renders are embedded as base64 data URIs so the file stays self-contained. Images are downscaled to 800px wide or less and skipped if the total exceeds 2MB; each image gets a one-line caption stating what to look at.

Can the HTML assembly be delegated to another model?▼

Yes, mechanical HTML assembly can be delegated to a grunt-tier model by handing it the template plus a structured outline of every section. Claude still writes the judgment sections and verifies the rendered result before delivery.

What happens if no renders changed during a run?▼

The visual evidence section is replaced with a single sentence explicitly stating that no renders changed. The skill prohibits padding the report with unrelated images or boilerplate content.