Extract post-merge learnings from pull requests into knowledge graph updates.
npx skills add https://github.com/agorokh/applied-ai-research --skill learner-agorokh
Please help me install this Agent Skill. Skill: learner Source: https://github.com/agorokh/applied-ai-research/tree/main/.cursor/skills/learner Command: npx skills add https://github.com/agorokh/applied-ai-research --skill learner-agorokh
After a PR is merged, the team often loses valuable learnings that could prevent repeated fixes and improve enterprise-readiness over time.
Run the learner skill after a PR is merged to extract post-merge learnings, update AGENTS.md learned workspace facts, and propose new or refined vault nodes based on the Tier-3 substrate query.
High-intent search queries and answers about installing and using this skill.
To document pull request learnings, analyze the merged PR diff and commit messages to propose AGENTS.md updates and schema-valid vault nodes. It queries the knowledge graph first to avoid duplicating prior learnings and tags facts as universal or project-specific.
Knowledge graph querying prevents duplicated documentation updates by requiring a mandatory Tier-3 substrate query before extraction. This grounds proposed vault node creation and AGENTS.md edits in existing canonical facts, ensuring new learnings complement rather than repeat prior knowledge.
Yes, you can use this for enterprise AI repository maintenance. It analyzes merged PR diffs to propose concise AGENTS.md learned facts updates and vault node refinements, summarizing proposed edits for human acceptance rather than silent replacement.
The learner skill requires the CLAUDE_MEMORY_GATE dependency to function. This prerequisite enables the mandatory Tier-3 knowledge graph queries needed to ground extracted learnings in existing canonical facts before proposing AGENTS.md or vault node updates.
Automated knowledge extraction from PR diffs has strict no-secrets constraints to prevent leakage. It focuses on summarizing proposed edits for human acceptance rather than silent replacement, meaning maintainers must review and approve all AGENTS.md and vault node updates.
The best way to maintain vault graph nodes is to run the learner skill post-merge to propose new or refined Tier-2 nodes. It applies schema-valid frontmatter and tags updates as universal versus project-specific based on extracted PR learnings.
ascorbicExtract actions, facts, and decisions from conversations into structured JSON.
pecorapromotionsllc-cellAudit local map presence and GBP health across platforms.
CHENyiru3Inspect markdown wiki and Obsidian vault schema before editing.
anicdhIndex repository knowledge graphs and generate CLAUDE.md and AGENTS.md context files.
