qte77 pushed to refs/heads/main at Lambda-Biolab/i3mega-pipettebot

Commits on May 8, 2026

1. chore(.claude): mirror compound-learning + context + core-principles …

   
   

   …rules into the repo (#20)
   
   The repo already ships motion-safety.md and pipette-delegation.md
   under .claude/rules/. This commit pins three more general-purpose
   agent rules (byte-identical to the maintainer's user-global versions)
   into the repo so any contributor running Claude Code in this tree
   gets the same governance even without the equivalent dotfiles in
   $HOME.
   
     - compound-learning.md   Promotion path for repeated mistakes:
                               1st occurrence -> fix inline,
                               2nd -> AGENT_LEARNINGS.md,
                               3rd -> .claude/rules/,
                               recurring workflow -> .claude/skills/.
     - context-management.md  ACE-FCA principles (target 40-60% capacity,
                               compaction triggers, subagent usage).
     - core-principles.md     KISS / DRY / YAGNI / "concise and focused"
                               decision framework, with the "STOP. Ask
                               the user." escape hatch.
   
   These are project instructions (not session-state) and supplement
   AGENTS.md. They are not consumed at runtime by `pipettebot` itself —
   Python code is unaffected. No CI impact.
   
   Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

   
   
   
   
   

   and
   claude

   authored

   May 8, 2026

   
   
   Configuration menu

   
   

   Browse the repository at this point in the history