mphaxise Learn
Use this skill when the repo should retain what it has learned instead of rediscovering the same patterns every time.
This port is adapted from garrytan/gstack at commit 4d2c8d94d00cc4f4f3d4c26316a4f939ceedc045.
Important Adaptation
Upstream gstack keeps a dedicated learnings layer. In this repo, the closest durable equivalents are:
brain-opsquerysignal-detectorreports
Workflow
- Find the existing learning or related context first.
- Add or update only durable patterns:
- preferences that repeat
- pitfalls that recur
- repo-specific decisions worth remembering
- Prune stale or contradicted learnings when the user asks.
- Export or summarize the current learnings cleanly when needed.
Guardrails
- Do not promote one-off observations into durable learnings too quickly.
- Keep the learning tied to evidence or repeated behavior.
- Prefer updating an existing learning over duplicating the same point.