harvest_skill

Overview

This skill harvests GitHub pull request data and turns it into clear reports: weekly/monthly summaries, release notes, client PDFs, and individual work reports. It uses the gh CLI to collect PR metadata, aggregates statistics, and formats outputs for stakeholders. Reports prioritize data quality, context, and reader-focused formatting.

How this skill works

The skill runs gh pr list with configurable filters (state, author, label, date range) to collect PR JSON, then applies incremental caching and jq-style filtering to minimize API load. Collected data is analyzed for metrics like additions/deletions, merge rate, review time, and cycle time. Final outputs are generated in Markdown, JSON, HTML, or PDF, with optional charts and member-level breakdowns.

When to use it

• Generate a weekly or monthly team PR summary for managers.
• Create release notes or a changelog for an upcoming deployment.
• Produce individual contribution reports for performance reviews or billing.
• Analyze PR quality trends and review time regressions.
• Export client-facing reports or PDFs with charts and work-hour estimates.

Best practices

• Confirm repository and filters before running large queries; ask for scope when >100 PRs.
• Choose output format early (Markdown/HTML/PDF/JSON) to shape analysis and visuals.
• Keep sensitive fields out of reports; redact or aggregate personal data.
• Use caching and incremental collection for repeated runs to avoid rate limits.
• Validate gh CLI auth and network connectivity; implement exponential backoff on errors.

Example use cases

• Weekly merged-PR summary with totals, top contributors, and cycle-time trends.
• Release notes generator that converts PR titles and descriptions into a changelog draft.
• Client report: HTML/PDF with charts, PR table, and estimated work hours per ticket.
• Individual work report listing a user’s PRs, time estimates, and review feedback.
• Quality trends report integrating Judge feedback to highlight recurring issues.