This captures recurring failure patterns from your CI runs and promotes them to durable prevention rules without needing interactive chat. It reads PR check results, deduplicates patterns by stable keys, and only suggests promotion after hitting recurrence thresholds (default: 3+ occurrences across 2+ runs in 30 days). The output is structured YAML you can feed into agent context files or workspace memory. Use it when you want headless learning loops in GitHub Actions pipelines. The recurrence filtering is the smart part here: it prevents one-off noise from polluting your system prompts while surfacing patterns that actually matter across your codebase.
npx -y skills add pskoett/pskoett-ai-skills --skill self-improvement-ci --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
sickn33/antigravity-awesome-skills
moizibnyousaf/ai-agent-skills
github/awesome-copilot