void-retrospective

Analyzes git history and PR data over a window to surface engineering signals and improvement decisions.

Updated May 29, 2026
One-click install
npx skills add https://github.com/voidcorp-core/void-harness --skill void-retrospective-voidcorp-core
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: void-retrospective
Source: https://github.com/voidcorp-core/void-harness/tree/main/packages/core/skills/void-retrospective
Command: npx skills add https://github.com/voidcorp-core/void-harness --skill void-retrospective-voidcorp-core

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Single lessons get captured in the moment, but patterns like recurring-fix files, fix-heavy commit mixes, and shrinking test-to-code ratios only become visible when you deliberately review a whole window of work. This Skill performs that periodic retrospective over git history, merged PRs, and skill-usage logs, turning raw history into concrete improvement decisions. ## Core Features & Use Cases - Signal computation: Parses conventional-commit types, file hotspots, recurring-fix files, test-to-production ratio, PR size distribution, and regressions from git log and GitHub PRs. - Decision-oriented report: Produces a window summary where every signal that warrants action maps to a concrete change such as a test to add, a file to refactor, or a convention to adopt. - Pattern routing: Routes durable, generalizable patterns to the companion lesson-capture skill for human-confirmed recording, while writing nothing itself. - Use Case: At the end of a sprint, ask for a retro since the last release and receive a report showing that one adapter file received three fix commits, the test ratio dropped, and PRs grew large, each paired with a specific remediation. ## Quick Start Ask the agent to run a retrospective over the last two weeks of git history and merged PRs and report the signals with improvement decisions.

Frequently Asked Questions about void-retrospective

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

FAQPage Schema
How do I run an engineering retrospective over git history?

Invoke the skill with a window such as a week or sprint, and it reads git log with authors, timestamps, and changed files, plus merged PRs via the GitHub CLI when a remote exists. It reports signals like hotspots and test ratio, ending in concrete improvement decisions.

What signals does a git history retrospective reveal?

It computes the conventional-commit type mix, most-changed hotspot files, files with repeated fix commits, test-to-production code ratio, PR size distribution, and regressions or reverts. Each signal is reported with evidence rather than a gamified score.

Does this retrospective require GitHub or specific tooling?

No. Git log is the only required source. PR data via the gh CLI and .void skill-usage logs are used when present, and the skill notes any missing source and proceeds with what exists.

Does the retrospective modify code or project documentation?

No. It only reads history and produces a report. Durable patterns are routed to the companion lesson-capture skill, which writes them only after human confirmation, so the retro itself never changes code or doctrine.

When should I use a window retrospective instead of capturing lessons as they happen?

Use point capture for a single lesson the moment it appears, and the window retrospective on a cadence to find patterns no single moment reveals, such as a file fixed three times or a steadily declining test ratio.