📖About

Third-attempt root fix of the team build lane (prior: 2972 flag-on, 2972r1 droid binding, 3055 Forge tools — lane still produces narration without work). Scope: (1) Sage fact pack and Archon design must exist as stored artifacts or the step fails early with an honest reason; (2) Argus unavailable or criterion-NOT-met blocks step completion (fail-closed); (3) step review grades stored evidence (artifacts + real file changes), never narration; (4) role Palantir posts carry evidence references; (5) meta root cause of the three failed fixes diagnosed and guarded by a regression test so a fourth silent regression is impossible.

⚖️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)
• META ROOT CAUSE of three failed team-lane fixes (2972, 2972r1, 3055): none was wrong — none ever RAN. Every live team run predates the fix commits (last run 2026-07-19 01:47 vs fixes 11:57 and 21:00); zero team runs since. The fixes were graded by source-string tests + mocked call shapes; the single live check skips itself when no data exists (forever); and the task-pipeline lane fell OPEN to the solo skill bridge on team failure, so a dead team looked green on every board. I verified my fixes by the same narration-shaped evidence the pipeline uses — the meta-bug and the object bug are the same bug: accepting description as proof of work. — Git history (fix commits 502afc1bf, b45f724b2 both AFTER the last team run), deliverable store contents (all rows are role-call-failed stubs), test sources (source-string asserts + skip-on-no-data), me_step_exec fallthrough handling only completed. (or-bx-01)
• BUILT (5 files + 1 test): (1) task-pipeline lane fail-CLOSED — team_failed or a team-lane crash fails the run honestly, no solo rebuild; solo runs only when the team never engaged, visibly logged. (2) Merlin refuses done without all four deliverables stored and non-degraded; done reasoning cites the stored versions. (3) Step Reviewer evidence gate: any team artifact present makes the full non-degraded chain + Argus criterion_met mandatory for a pass, read from the store never from posts. (4) Deliverable persistence no longer fail-silent; role posts carry artifact references as evidence; Merlin posts a visible refusal when evidence is incomplete. (5) Usage capture restored through call_llm (severed 2026-07-09) so team roles reappear in cost surfaces. (6) test_ini3084: 22 checks driving the real code paths hermetically, never skips — registered in regression. Cookbook 863 (the law) + 864 (the meta-lesson). — All 22 contract checks pass on the deployed code; adjacent tests test_ini3055 (9 pass) and test_ini2972_r1 (9 pass) stay green. (or-bx-01)
• [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

• A team build step never shows completed unless Sage fact pack and Archon design are stored and Forge made real file changes (must_have)
• A step whose test verdict is missing, errored, or NOT met never shows completed (must_have)
• Ninad can see per step WHICH artifacts each agent delivered, from the stored evidence itself (must_have)
• The reason three prior fixes did not stick is documented on this initiative and a regression test fails if the team contract breaks again (must_have)