Child of #1058, item 6. Design work: an agent that needs unconfigured engagement inputs still runs, consumes a run and a budget slot, then reports not evaluable.
Expected: unmet configuration prerequisites are shown on Overview and Configure before Run, with a per-check needs list. Block the run only when every check is unrunnable. Label a config-driven partial distinctly from an envelope-driven partial.
Actual: structural prerequisites are enforced at install (assert_installable), but no config-prerequisite metadata exists anywhere in the bundle manifest chain (persona store, admin agents_bundles, push payload, bench listing). The gap only surfaces post-run via the configuration_missing coverage reason.
Suspected location: new manifest surface across jarvis-persona bundles, jarvis_admin agents_bundles.py, and jarvis/chat/agents_api.py.
This issue starts as a design document; implementation will be scoped from it.
Child of #1058, item 6. Design work: an agent that needs unconfigured engagement inputs still runs, consumes a run and a budget slot, then reports not evaluable.
Expected: unmet configuration prerequisites are shown on Overview and Configure before Run, with a per-check needs list. Block the run only when every check is unrunnable. Label a config-driven partial distinctly from an envelope-driven partial.
Actual: structural prerequisites are enforced at install (
assert_installable), but no config-prerequisite metadata exists anywhere in the bundle manifest chain (persona store, admin agents_bundles, push payload, bench listing). The gap only surfaces post-run via the configuration_missing coverage reason.Suspected location: new manifest surface across jarvis-persona bundles, jarvis_admin agents_bundles.py, and
jarvis/chat/agents_api.py.This issue starts as a design document; implementation will be scoped from it.