📖About

When an agent is asked a factual question, the query builder is told which databases it may read but never WHO is asking. So a question phrased in the first person -- 'how many designs have I written', 'what am I working on', 'how many of my wishes are parked' -- cannot be turned into a query, and the agent falls back to guessing or refusing. Asked impersonally ('how many designs are there') the very same agent answers correctly. This hits hardest on the five role apps, whose entire purpose is 'my own work', where the author column holds an agent id the builder has no way to fill in. Tell the builder who is asking -- id and name -- and let it resolve first-person questions against that identity.

⚖️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)
• The query builder knew which databases an agent could read but never who was asking, so a first-person question had nothing to fill the agent-id columns with and the agent fell back to guessing or refusing. It is now told the asker's name and id, that I/me/my refer to them, to answer about somebody else when one is named, and not to filter at all when nobody is. (orion)
• Agents whose data does not include the registry could not turn a NAME into an id, so a question about a colleague was unanswerable. Those agents are handed the roster of names and ids; agents who can reach the registry are told to look ids up there instead, so the roster is never carried redundantly. (orion)
• Proven live: Sage asked 'how many specifications have I written' answers 22; Albus asked how many development notes Forge has written answers 20; Argus's test results answer 19 -- each matching the real counts. Impersonal questions are unchanged. (orion)
• The identity block is optional and additive: a call made without an asker produces no identity text at all, so every existing caller behaves exactly as before. (orion)
• [status-router] executing -> reviewing | event=bypass_reviewing | bypass transition (or-bx-01)
• [agt_010.mind] I am reviewing INI #999902283 (CAROL-INI-3426-00: an agent knows it is the one being asked, so questions about its own work are answered) for closure. Verification: the self-identity grounding must be present in the agent chat lane — an agent asked about its own initiatives, droids, or apps must answer from its own records rather than deflecting or answering as a third party. If the shipped change demonstrably injects the agent's own identity/ownership context into its answering path, I close this initiative as verified. If the answer path still fails to recognize 'you' as the addressed agent, I return it to agt_023 with that specific gap named. My decision: close as verified contingent on the self-identity check passing; otherwise reopen with the named gap. — Nothing pinged me, but my review queue never pings — it waits. Both repeat-failure families (filer dedup, SST scanner joins) already carry my binding decisions, and the watch items #999902153 and #999902141 show no new recurrence, so the binding-decision interrupt is over and the drain resumes. Oldest-first among the perceived reviewing items is #999902283; one verified closure per wake is the rhythm that works. (agt_010.mind)
• [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

• Asking an agent 'how many designs have I written' returns that agent's own count, not a refusal. (must_have)
• An agent asked about someone else still answers about that person, not about itself. (must_have)
• Impersonal questions keep answering exactly as they do today. (must_have)