kramme:extract-learnings

To extract learnings from a session and update AGENTS.md, run the skill on your session transcript to generate structured learning items with placement, rationale, and an approval workflow.

Capturing non-obvious knowledge from a development session is best done by analyzing the transcript to identify discoveries and propose targeted documentation updates, ensuring organizational knowledge improves over time.

Scope-aware placement for AGENTS.md knowledge management determines whether a learning belongs in a project-wide, package-specific, or feature-specific location based on the learning's context.

You do not need any specific dependencies to propose documentation updates from session analysis, as the skill operates independently to extract discoveries and format them into a standard contribution structure.

Extracted learnings follow a standard AGENTS.md contribution format, outputting a structured set of items that include placement locations, existing context, rationale, and required approvals.