ai-learnlisted

# Learn ## Purpose Continuous improvement from delivery outcomes. Analyzes merged PRs to find where AI missed what human reviewers caught, identifies false positives, and writes lessons directly to `.ai-engineering/LESSONS.md`. The feedback loop that makes the framework smarter over time. ## Trigger - Command: `/ai-learn single <pr>|batch` - Context: after PR merge (single), periodic review (batch). Step 0: read `.ai-engineering/LESSONS.md` for pre-existing patterns; load stack contexts: read `.ai-engineering/manifest.yml` `providers.stacks` and apply `.ai-engineering/overrides/<stack>/conventions.md` for each stack. ## Workflow Two modes: `single <pr>` (analyze one PR) and `batch` (analyze all merged PRs since last lesson update). Both follow the same loop: 1. Read PR review comments + code-change diff. 2. For each comment, classify the lesson category (Pattern Categories below). 3. Check for duplicates against existing LESSONS.md entries. 4. Append new lessons with category + evidence link. 5. When enough lessons accumulate per category, optionally draft an AGENTS.md proposal. ## Modes ### single <pr> -- Analyze one merged PR 1. **Fetch PR data** -- `gh pr view <pr> --json body,reviews,comments,files,additions,deletions`. 2. **Collect AI findings** -- read the AI-generated PR description, guard advisories, and verify results from the PR. 3. **Collect human feedback** -- extract all review comments, requested changes, and approval notes. 4. **Cross-reference** --