feat(policy): agent-driven policy management — the agent half#1323

Merged

zredlined merged 16 commits into

mainfrom

feat/agent-policy-loop-v2

May 13, 2026

Commits on May 13, 2026

feat(policy): plumb chunk_ids and rejection_reason through proposal pipeline
zredlined
committed
feat(sandbox): add /v1/proposals/{id} and /wait long-poll to policy.local
zredlined
committed
docs(sandbox): teach policy_advisor skill the wait + redraft loop
zredlined
committed
test(policy-advisor): add end-to-end smoke for the agent feedback loop
zredlined
committed
fix(policy-advisor): surface real CLI errors from wait-smoke preflight
zredlined
committed
fix(policy-advisor): pass --json to settings get in wait-smoke preflight
zredlined
committed
fix(policy): dedup draft chunks only in mechanistic mode; return effective id
zredlined
committed
docs(examples): retarget policy-management demo at the /wait endpoint
zredlined
committed
fix(examples): pass --yes on demo.sh's global setting writes
zredlined
committed
feat(sandbox): emit OCSF audit events for policy proposal lifecycle
zredlined
committed
feat(policy): /wait awaits local policy reload; demo auto-approves redrafts
zredlined
committed
fix(sandbox): /wait detects policy reload by content, not the schema version
zredlined
committed
fix(examples): redact tokens with python literal-string replace, not sed
zredlined
committed
fix(sandbox): scope /wait reload check to the approved rule
zredlined
committed
fix(server): make GetDraftPolicy dual-auth so /wait works under OIDC
zredlined
committed
fix(ci): address rebased check failures
zredlined
committed