📖About

Ninad ruling (2026-08-14, this session): work raised by an agent's escalation is UAT'd by that agent when it next wakes - the operator must ensure the agent is INSTRUCTED and ENABLED to do it. Applied first to Clara's two escalations (Agentbook 44/45 -> initiatives 3745 and 3836).

Mechanics, all record-driven, no protected-core change: (1) the UAT acceptor is already DERIVED from the record (3518: whoever requested does the UAT), so the filings are re-homed to the raising agent via the 3746 identity-correction lane - requester clara, owner agt_010 - and the derivation lands on her with zero code change. (2) The wake gains a uat_verdict action: enum + instruction line in the deliberation contract, an executor that requires verdict AND first-person evidence, posts the evidence as a UAT decision row, then signs via the initiatives app UAT endpoint with the acting agent's own registry name (server-side acceptor validation makes it spoof-proof). (3) uat_verdict is WORK (a pass closes an initiative) so it is conscious-action-whitelist gated (P.01.03.01.09, fail-closed); the operator widens it for agt_010 only. (4) The wake perceives an awaiting_my_signoff block: the agent's own uat-pending initiatives with the latest operator-written UAT instructions and the live last result of the initiative's regression test from the persistent runner store - so re-performance evidence is in front of her at wake time. (5) Agentbook replies to her escalations carry the same instructions.

⚖️Decisions

• Elrond's bypass methodology checklist (a reminder, not a gate -- you've got this): 0. File it requested_mode='bypass' (planner-vs-bypass is a deliberate choice). bypass_start REFUSES a non-bypass initiative (CAROL-INI-1846), and the dispatcher only skips the bypass lane when the mode says bypass -- a 'planner' mistag lets Merlin's pipeline grab the placeholder step and block your finished work. 1. Filed as planned status -- let the bypass claim/activate it; never file active. 2. Open the bypass (bypass_start) with your droid id + the remediation answer (remediates_initiative_id=NNN, or remediates_nothing=True). 3. Work the blocks for your work-type: template -> design -> code -> test -> review. Do the real work; record decisions on the initiative as you make them. 4. Reality is recorded for you at close -- code (files changed), each decision, and the twin-review verdict become real activities tied to this initiative and show in the Activity Tracker like a planner run (CAROL-INI-1840). No dummy rows. 5. Keep the initiative status moving; it parks in 'reviewing' and is tagged uat-pending for you at close (CAROL-INI-1836), so the stuck-watchdog leaves it alone until UAT. 6. Close runs the gates (design/architecture compliance + caller-audit). If a gate flags something pre-existing or unrelated to your change, waive it with a clear written rationale -- audit, don't skip. 7. Bypass skips the planner's auto-orchestration, NOT the standards. Same template checklist, same review, same observability as a planner run. (elrond)
• [status-router] planned -> executing | event=bypass_executing | bypass transition (or-bx-01)
• [delivery-check] 6 pending must-have criteria stamped met at bypass_end on live re-performance evidence (CAROL-INI-3020): test test_ini3837.py: PASS (7 passed in 0.36s) (orion)
• [status-router] executing -> reviewing | event=bypass_reviewing | bypass transition (or-bx-01)
• [status-router] reviewing -> closed | event=operator_signoff | Auto-accepted (CAROL-INI-1859): Orion-initiated, >2 days in reviewing with no objection. (el-srac-01)

✅Success criteria

• acceptor derivation lands on the raising agent for both re-homed filings (3745, 3836): acceptor_for returns clara/agent for each, proven by query, with owner-equals-requester preserved (clara / agt_010). (must_have)
• Clara's wake offers uat_verdict: the action is in the deliberation enum with an instruction line requiring first-person evidence; the executor refuses a verdict without evidence and refuses an agent who is not the derived acceptor (server-side check). (must_have)
• uat_verdict is whitelist-gated work: with no whitelist row it is REFUSED (fail-closed proven); with the operator-added agt_010 row it executes for Clara only. (must_have)
• the wake snapshot for Clara contains awaiting_my_signoff listing her uat-pending initiatives with operator UAT instructions and the live latest regression-test result for each. (must_have)
• a regression test in the persistent runner locks all of the above (enum, fail-closed refusal, whitelist row, acceptor derivation, perceive block) and is baselined with zero NEW FAILURE. (must_have)
• the ruling is promoted to the cookbook in the same session and cited by number in the handover. (must_have)