{% extends "base.html" %} {% block title %}Proposals{% endblock %} {% block topbar_main %}
Learning is scoped to a repository, so this page needs one. Evidence from one repository proposing a skill in another is exactly the leak the scoping exists to prevent.
Nothing is harvested. Watching skill use sees names and times, not what you were working on, and nothing reads your conversation history. A proposal exists because somebody handed over the work it came from.
Approving covers the exact draft you read. Editing it afterwards sends it back for another look rather than shipping the edit under the old approval.
Nothing proposed. flanner skills evidence submit hands
over a piece of work, and flanner skills propose drafts
a skill from it.
{{ open.draft_hash }}
{% if open.rationale %}{{ open.rationale }}
{% endif %}No evidence is readable. It may have expired.
{% endif %} {% if open.evidence_expired %}{{ open.evidence_expired }} piece(s) have since expired.
{% endif %}{{ note }}
{% endif %}{% endfor %}{{ arm }}: {{ stats.passed }}/{{ stats.trials }} passed ({{ stats.reads_as }}), {{ stats.not_run }} cell(s) not run.
{% endfor %} {% for limit in matrix.limits %}{{ limit }}
{% endfor %} {% for note in matrix.notes %}{{ note }}
{% endfor %}Flanner records comparisons; it does not run them. Nothing here calls
a model provider, so results are entered by whatever actually ran
them โ flanner skills eval record.