✔ fixed normal exploration, pagination, SHA changes and repeated ranges are external observations, not a score (1.3485ms) ✔ read-only research is valid work; model summaries and unchanged timestamped listings do not become engineering evidence (0.371ms) ✔ new permissions, network and broken tools are failure evidence; repeated invalid inputs remain candidates (0.3971ms) ✔ waiting for original long checks, review and approvals stays waiting across repeated observations (0.4962ms) ✔ baseline facts, real repairs, current-source check results and requirement evidence have distinct growth (0.2112ms) ✔ bounded memory keeps unknown instead of claiming an unretained range is repeated; Task/Run instances cannot mix (12.5194ms) ✔ actual runtime records growing pages then a repeated page without changing model rounds, stop rules or original tool pairs (3100.7065ms) ✔ actual invalid tool calls repeat under the original step budget; observation never adds an early stop (2435.2265ms) ✔ actual command approval is observed as waiting and denial stays a permission failure (1991.3118ms) ✔ network errors remain model failures with shadow failure evidence, never a fabricated no-progress cause (3501.8394ms) ✔ observer receipt failure cannot change tool results, model budget or completion (1567.6045ms) ✔ a later receipt recovers after an observation write failure without misaligning retained source receipts (2294.3541ms) ℹ tests 12 ℹ suites 0 ℹ pass 12 ℹ fail 0 ℹ cancelled 0 ℹ skipped 0 ℹ todo 0 ℹ duration_ms 15107.0841