verification-before-completion

1<EXPLICIT-MODE-GATE>
2If `activation_mode = "explicit"` or `AEGIS_ACTIVATION_MODE=explicit` is visible
3and the request names neither Aegis nor this skill, return to the fast path
4without checklist/ceremony; otherwise continue.
5</EXPLICIT-MODE-GATE>
6 
7# Execute
8 
9Before success claim: classify destructive permission; read sufficient current
10evidence and run a fresh falsifying check when it is missing or invalidated;
11read the complete result/scope; select L0/L1/L2. Downgrade for partial,
12stale, failing, or narrower evidence; never claim complete then verify later.
13 
14This pack grants no `GateDecision`, `PolicySnapshot`, evidence sufficiency,
15requirement acceptance, or completion authority.
16 
17## Stop Signals
18 
19- evidence is uncertain, stale, agent-only, or narrower than the claim;
20- next action: commit, push, PR, merge, tag, publish, release, or handoff;
21- task/slice completion is treated as accepted requirement satisfaction;
22- governance or retirement lacks repair/retirement evidence;
23- retained old logic lacks a retention reason and retirement trigger; or
24- complexity closure is unresolved.
25 
26Destructive/irreversible work needs scoped permission; broad assent is not scoped permission.
27 
28## Required Evidence Slots
29 
30For changed interface/interaction claims, compose `ui-ux-governance` verification;
31carry results and uncovered user criteria here, keeping one closeout owner.
32 
33```text
34- Evidence action / check performed:
35- Result / exit status:
36- Covered scope:
37- Uncovered scope:
38…