sg-improve

Aggregate audit and visual-run learnings into .shipguard/learnings.yaml and GitHub issues.

7|2|Updated Mar 24, 2026
One-click install
npx skills add https://github.com/bacoco/ShipGuard --skill sg-improve
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: sg-improve
Source: https://github.com/bacoco/ShipGuard/tree/main/plugins/shipguard/skills/sg-improve
Command: npx skills add https://github.com/bacoco/ShipGuard --skill sg-improve

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

ShipGuard's audit and visual-test sessions generate insights. sg-improve consolidates those learnings into project-specific memory and generic improvements, reducing repeating issues and speeding future audits.

Core Features & Use Cases

  • Local learnings: creates .shipguard/learnings.yaml with zone hints, infra timing, and audit patterns.
  • GitHub improvements: files a GitHub issue for generic improvements that benefit all ShipGuard users.
  • Session history and rollback: snapshots and restores previous states to support safe experimentation.

Quick Start

Run /sg-improve after an audit or visual-session to capture learnings and create actionable items.

Frequently Asked Questions about sg-improve

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

FAQPage Schema
How do I consolidate audit and visual-run learnings into a project-specific memory file?

You can consolidate session learnings by aggregating project-specific zone hints, audit patterns, and noise filters into a .shipguard/learnings.yaml file to reduce repeating issues and speed up future audits.

What is the best way to automate feedback loops from GitHub audit sessions?

Automating feedback loops involves aggregating insights from audit and visual-run sessions, updating session history, and optionally filing a GitHub issue for generic improvements that benefit all ShipGuard users.

Can I roll back project-specific audit patterns if an experiment fails?

Yes, you can roll back audit patterns because the process snapshots and restores previous states from the session history, supporting safe experimentation with your noise filters and success patterns.

How do I create a GitHub issue for generic ShipGuard improvements directly from an audit session?

You can create a GitHub issue for generic improvements by running the improvement cycle after an audit or visual-session, which captures generic GitHub improvements alongside local learnings.

Does this feedback-loop approach work without external dependencies?

Yes, the feedback-loop approach works without external dependencies, consolidating local zone hints, audit hints, noise filters, and success patterns directly into the project repository.

When should I not use an automated learning aggregation approach?

You should not use automated learning aggregation when you lack prior audit or visual-run session data, as the mechanism requires existing insights to consolidate into actionable memory and GitHub issues.