What problem does it solve? Teams repeatedly re-derive the same fixes because bugs, wrong assumptions, and design trade-offs are never captured in a searchable form. This Skill enforces a structured learning log so every non-trivial problem, its root cause, candidate solutions, and generalizable lessons are recorded and searchable before the next occurrence. ## Core Features & Use Cases - Structured problem tracking: Records problems, root causes, candidate solutions (including rejected ones), accepted fixes, and generalized learnings in a local SQLite database. - Hybrid search before fixing: Searches prior records by keyword so recurring issues surface their accepted solutions instead of being re-derived from scratch. - Expiring observations: Captures later refinements, recurrences, and corrections as nestable observations with TTL-based expiry and stale-flagging. - Use Case: Before fixing a flaky test that smells familiar, run a search to find a prior record showing the root cause was a shared fixture, then apply the already-accepted solution and log a new observation. ## Quick Start Ask the agent to search the learning log for prior occurrences of the current problem, then record the problem, its cause, and the accepted solution using the ether learn commands.
