self-improvement

Turn every captured lesson into a skill — extend an existing skill whose trigger already covers it, or write a new one. Use after a user correction, after the user states how they want a task done or how work is done in this repo, or after you learn something non-obvious from docs during the task.

Consolidate the accumulated skill corpus — global (~/.claude/skills) and project (.claude/skills). Run periodically to prune dead rules, unite overlapping ones, and generalize the over-specific. The complement to capture-lesson: that skill captures, this one consolidates.

Report the session archive the self-improvement scan has collected — what the backlog is made of, which clusters it falls into, and what /dream would work on next. Use when the user asks what lessons the scan stored, what the statusline lesson count is made of, or which lessons are worth harvesting. Building a course or study guide out of a document corpus is the loose `lessons` skill, not this one.

Use when a golden / snapshot / approval / fixture-capture file moves as part of a change and is about to be regenerated or accepted — `--snapshot-update`, `-update`, `pytest --force-regen`, `jest -u`, `cargo insta accept`, or a hand-run "regenerate the expected output" script. Also fires when reviewing a diff that contains both source edits and a regenerated capture. Regenerating makes the test green whether the movement was intended or not, so each moved field needs a cause named in the task before the new capture is accepted — and a capture cited as evidence for a case needs a row that actually exercises that case.