Learning

Audit autonomous memory and skill improvement, evaluations, canaries, and rollbacks.

Evidence
0
Queued reviews
0
Active skills
0
Passed evaluations
0
Autonomous Learning
Reviewer: gpt-oss:20b. Changes are limited to memories and user-created skill versions.
Worker idle0 queued0 running0 processedLast trigger: learning-page

Learned Skills

No learned skills yet
Successful repeated workflows and corrections will create evaluated skill candidates.

Skill Catalog

No catalog templates found
Bundled skill templates will appear here when available.

Recent Evidence

No learning evidence yet
Chat outcomes, feedback, tasks, tools, and skill runs will appear here.
Evaluation Policy
Direct corrections promote immediately. Inferred memories require corroboration and confidence of at least 0.80. Skills require a passing evaluation and three successful canary uses; failures automatically roll back.