extract-from-activity

Extract reusable lessons from repository commits, merged PRs, and issue activity.

Updated Jun 12, 2026
One-click install
npx skills add https://github.com/missingbulb/GoogleCalendarEventCreator --skill extract-from-activity-missingbulb
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: extract-from-activity
Source: https://github.com/missingbulb/GoogleCalendarEventCreator/tree/main/.claudinite/shared/packs/claudinite-growth/skills/extract-from-activity
Command: npx skills add https://github.com/missingbulb/GoogleCalendarEventCreator --skill extract-from-activity-missingbulb

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Valuable engineering lessons from a stretch of work — why a fix was shaped a certain way, what a reviewer pushed back on, why an approach was reverted — stay buried in commit history and PR threads instead of becoming reusable guidance for the repository. ## Core Features & Use Cases - Artifact Mining: Reads a defined window of commits (full bodies and diffs), merged PRs (diffs plus review discussion), and related issue comments to surface durable lessons. - Lesson Quality Bar: Filters out already-enforced implementation choices and keeps only lessons that steer future decisions or catch repeatable mistakes. - Local Pack Landing: Routes each accepted lesson down a promotion ladder — declared check first, custom code rule second, prose last — into the repo's own local packs, with red-first fixtures and since dates for new checks. - Use Case: After a two-week sprint, run this Skill over the merged PRs to capture review verdicts and non-obvious fix rationales as checks or pack entries before that context is forgotten. ## Quick Start Extract lessons from the commits and merged PRs of the last two weeks and land them in this repo's local packs.

Frequently Asked Questions about extract-from-activity

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

FAQPage Schema
How do I extract lessons from recent commits and pull requests?▼

Define a time window, then read the commits with full bodies and diffs, the PRs merged in that window including review discussion, and related issue comments. Convert signals like non-obvious fixes and review verdicts into lessons landed in the repo's local packs.

What makes a good lesson from code review activity?▼

A good lesson steers a future decision or catches a repeatable mistake, such as a convention a reviewer named or a reverted approach. An already-settled, test-enforced implementation choice is not a lesson and should be dropped.

Can this analyze conversation logs or meeting discussions?▼

No, this Skill reads only committed artifacts: commits, merged PRs, and issue activity. Dialogue-based signals like wall-time friction or clarifying round-trips belong to the sibling extract-from-conversations Skill working from logs.

Where do extracted lessons get stored in a repository?▼

Lessons land in the repo's own local packs, routed down a promotion ladder: a declared check first, a custom code rule where patterns cannot express it, and prose last. New checks ship with red-first fixtures and a since date.

What if a work window contains no extractable lessons?▼

Finding nothing is a valid and common outcome. Adding a duplicate or invented lesson is considered worse than adding nothing, so the run simply records no additions.