✔ engineering inspection preserves a dirty Git baseline and treats project rules as data (2702.5765ms) ✔ real npm tests and build produce counted results, log artifacts and persistent verification (2569.4318ms) ✔ an actual failing test retains the nonzero exit code and failed count (2252.8093ms) ✔ unchanged automatic checks return original failure evidence, while explicit user reruns and changed source still execute (3784.8343ms) ✔ agent loops cannot hide identical failing checks behind new plan IDs or unrelated reads (9510.7716ms) ✔ zero tests, skipped tests and unrecognized stdout cannot become verification passes (7995.0283ms) ✔ a successful build with no selected tests does not verify the task (2043.1069ms) ✔ stale check plans and forged script names execute no project commands (1539.6436ms) ✔ applying a later proposal invalidates the prior passed engineering snapshot (2793.5203ms) ✔ stopping an actual check terminates its child process and blocks concurrent jobs (2048.8725ms) ✔ check timeout records a failure with its actual termination result (2788.2215ms) ✔ source edits during a successful check make the result stale (3815.2259ms) ✔ bounded large logs retain the test footer and readable artifact instead of inventing counts (3024.1205ms) ✔ model-selected checks require a run-bound decision and denial creates no check job (4449.4773ms) ✔ check APIs expose owned plans and jobs and preserve the authoritative results across restart (7494.451ms) ✔ reviewed changes continue the original task without reopening a generic completed run (4759.3128ms) ✔ concurrent review-continuation API requests accept one run and retain the goal after restart (16309.1948ms) ✔ the model reads owned failure logs without running the check again (5343.3675ms) ✔ process exit at check queued preserves results and never replays commands (2216.8371ms) ✔ process exit at check result_committed preserves results and never replays commands (2870.5215ms) ✔ HTTP download preserves original malformed SSE bytes and artifact SHA instead of decoding invalid UTF8 (2867.0308ms) ✔ terminated SSE adapts empty optional tool arguments, retains the exact wire bytes and executes the validated batch once (3445.4665ms) ✔ reported SSE usage survives rejected required arguments "" without dispatching any tool (1483.337ms) ✔ reported SSE usage survives rejected required arguments "{\"path\":" without dispatching any tool (1591.5308ms) ✔ reported SSE usage survives rejected required arguments "null" without dispatching any tool (1457.1147ms) ✔ a stream without DONE retains reported usage but cannot adapt or execute even an optional tool (2224.6439ms) ✔ a timeout after reasoning-only fragments creates one unknown receipt and never invents reported tokens (2836.847ms) ✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (3710.1718ms) ✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (3583.5143ms) ✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (2475.7553ms) ✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (1879.7497ms) ✔ explicit streaming probe uses actual SSE despite unknown declaration and binds its own original attempt (982.1347ms) ✔ coalesced initial SSE events publish the latest partial state even when the upstream stays idle (2391.6893ms) ✔ stream timeout closes the original HTTP response, preserves redacted capture and never retries (1365.9387ms) ✔ SSE accepts CR-only final event boundaries and retains exact invalid UTF8 bytes without dispatch (0.7314ms) ▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data ✔ reject 0 (0.351ms) ✔ reject 1 (0.1877ms) ✔ reject 2 (0.1636ms) ✔ reject 3 (4.9466ms) ✔ reject 4 (0.1718ms) ✔ reject 5 (0.137ms) ✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (6.6902ms) ✔ continuation reads advance through historical pages without a false pause while repeated current-Run pages still pause (2.6777ms) ✔ a new tool page replaces old latest-feedback space and still obeys its escaped protocol budget (0.6335ms) ✔ budgeted artifact pages count escaped protocol bytes and reconstruct original multilingual JSON without losing characters (9.0948ms) ✔ known artifact coverage detects semantic repeats despite changing limits, and advancing pages or direct evidence restore progress (0.3422ms) ✔ actual large artifact reads are present in the next bounded HTTP request and next_offset progresses without redoing tools (4933.5479ms) ✔ a real repeated artifact loop pauses the same Task after recorded results, survives reload and only resumes explicitly (7328.3306ms) ✔ actual HTTP continuation after reload sends distinct historically covered pages without false no_progress or replay (8212.053ms) ✔ operational questions are diagnostic while explicit repair requests remain execution instructions (1.0844ms) ✔ HTML without tests ends at reviewed file delivery and never offers automatic rechecking (4032.3353ms) ✔ diagnostic answers retain the unfinished task, limits and check result, and enforce read-only tools (5030.3006ms) ✔ review continuation is a system action and adds no testing demand or fake user instruction (4684.2044ms) ✔ a provider cannot execute tests against an explicit no-test task even if it ignores advertised tools (2702.4452ms) ✔ alternating duplicate file reads and listing files still pauses without treating new pages as duplicates (24905.3452ms) ℹ tests 55 ℹ suites 0 ℹ pass 55 ℹ fail 0 ℹ cancelled 0 ℹ skipped 0 ℹ todo 0 ℹ duration_ms 94531.2905