evolve-suggest

You automate skill evolution proposals by running a process at session-close that aggregates pending feedback into per-skill value updates and queues concrete SKILL.md edits for approval. It consolidates feedback-log data to update scores and prepare proposals automatically.

Consolidating feedback into skill value updates involves aggregating pending G2 feedback to compute net deltas and dominant patterns across skills. This process updates per-skill scores, maintains session history, and recomputes the overall governance value.

Proposing edits to a skill configuration file based on feedback requires aggregating session data until improvement signals reach defined thresholds. Once signals are strong enough, targeted SKILL.md edits are queued as proposals rather than auto-applied.

No, the skill governance update process never auto-applies changes to configurations. It queues evolution proposals for explicit user approval via an apply step, ensuring you maintain control over any modifications to your skill definitions.

You need to run feedback aggregation for skill management at the end of a session, specifically after memory-write and memory-consolidate processes. This timing ensures all session feedback is captured to prepare evolution proposals for skills with actionables.

Yes, you can use session feedback to recompute the overall governance value. The aggregation process consolidates feedback across skills to compute net deltas, updates per-skill scores, and recomputes the global V^L value function.