v0.2.0 — slash commands

Added

• Three on-demand slash commands for the Claude Code plugin — invoked manually, nothing fires automatically:
  • /learned-behavior:advice — show approved lessons for the current workspace
  • /learned-behavior:review — summary of stored lessons and recurring errors for this workspace
  • /learned-behavior:mine — run the four miners (errors, edits, skill bypasses, guard blocks) and surface new candidate lessons
• README: badges, "What it isn't" section, end-to-end example, troubleshooting
• CI: pytest matrix on Python 3.10/3.11/3.12 + plugin manifest validation
• PyPI release pipeline via GitHub Actions Trusted Publishing (no stored token)

Notes

• Promotion of mined candidates remains a deliberate, separate step (learning.py promote --write); /learned-behavior:mine only proposes.

Full changelog: https://github.com/lisn0/learned-behavior/blob/main/CHANGELOG.md