reflect
You should use this skill when: a task is completed, a session is long: its context was compacted, it made dozens of tool calls, the agent started repeating earlier steps, or more than half of the context window is used when a token counter is visible, or when a scheduled or manual review of recent agent sessions is requested, to reflect on the sessions, propose the 3 most important improvements to AI context files that would improve speed or quality next time, and record the run in a journal of reviewed sessions and decisions
Pinned to revision 3b6136dc9e4d, so it is the text this page describes rather than whatever the author pushed since.
Files
Every link opens the file at its source, pinned to the revision this page describes.