
> tablaze@0.1.0 test
> npm run build && node --test --test-concurrency=1 tests/*.test.mjs


> tablaze@0.1.0 build
> tsc -p tsconfig.json

✔ workspace export keeps active selection consistent when the active tab closes during storage capture (1173.805375ms)
✔ PDF metadata stays bound to its captured tab through asynchronous artifact persistence (705.611209ms)
✔ drag revalidates observed targets after asynchronous geometry reads before mouse input (673.814959ms)
✔ upload chooser revalidates its visible trigger after asynchronous local file checks (593.863334ms)
✔ newly discovered scroll containers retain node identity guards when replaced (503.04375ms)
✔ scrollable containers expose ranges and accept guarded horizontal and vertical input (605.0545ms)
✔ dragging emits the browser drop event and delivers the real data payload (676.369ms)
✔ hidden file chooser transfers explicit local bytes without a hidden DOM reference (720.196042ms)
✔ PDF exports real private artifacts with verifiable digest and source (622.438834ms)
✔ workspace roundtrip restores isolated durable storage, owned tabs, and active selection (1920.198666ms)
✔ workspace validation precedes navigation and failed restore rolls back owned sessions (873.259375ms)
✔ per-decision catalogs hide execution metadata and context changes skip queued writes (12.742042ms)
✔ a context change during planning or write-ahead persistence never executes the stale handler (2.644375ms)
✔ trusted write effects retain ambiguity for thrown, malformed, error, and unknown outcomes (7.911833ms)
✔ finish rechecks its lease and cannot reuse verification from the old context (1.91275ms)
✔ a context change between closed and newly prepared leases invalidates old verification (3.528125ms)
✔ only a confirmed session close preserves current evidence across context changes and later writes still invalidate it (4.56975ms)
✔ bound checkpoints reject missing runtimes and changed registry or same-origin tenant identities (1.131416ms)
✔ version-one and version-two checkpoints explicitly migrate unbound and cannot gain a registry (1.291417ms)
✔ resume rechecks a pending call effect rather than trusting a tampered mutating flag (1.109917ms)
✔ initialization uses the bound dispatcher and each lease closes before terminal persistence (1.605041ms)
✔ a prepared lease that arrives after cancellation is still closed without planning (2.977416ms)
✔ catalog release failure or cancellation cannot turn accepted verification into success (2.0855ms)
✔ run CLI completes a scripted model task through real Chrome and prints a bounded report (898.093042ms)
✔ run CLI returns human input with exit 2 and an explicit failure with exit 1 (418.992042ms)
✔ run CLI does not accept an unverified model completion and enforces the step limit (225.55075ms)
✔ CLI rejects agent-only options outside run and validates required arguments and limits (2316.839417ms)
✔ CLI reports cleanup failure separately from an already verified successful agent result (719.043209ms)
✔ restoration abort handles a rejected asynchronous dispose without an unhandled rejection (324.775459ms)
✔ CLI publishes bounded planner diagnostics without exposing endpoint, key, or upstream error text (216.9705ms)
✔ HTTP planner failures expose only status and fixed classification without enabling default retries (19.854041ms)
✔ unknown exceptions cannot inject diagnostic codes and custom retry policy remains authoritative (2.772542ms)
✔ transport and malformed response diagnostics remain distinct and omit raw response data (3.266292ms)
✔ catalog failure is distinguishable before any planner or mutation runs (1.953166ms)
✔ application callbacks are classified without recursively notifying a broken event hook (5.181083ms)
✔ adapter usage callback failures retain their application phase and original retry policy input (1.162041ms)
✔ persistence failures prevent mutation or successful completion without changing resume schema (2.15525ms)
✔ a noncooperating checkpoint callback receives a bounded persistence-timeout diagnosis (5002.926625ms)
✔ ordinary MCP explicit unknown write persists uncertainty, skips the batch and cannot replay on resume (17.6545ms)
▶ explicit uncertainty overrides a superficially successful write in legacy and bound MCP execution
  ✔ ordinary MCP write annotation (1.59375ms)
  ✔ bound trusted write metadata (2.579958ms)
✔ explicit uncertainty overrides a superficially successful write in legacy and bound MCP execution (4.376291ms)
▶ ordinary pre-write failures and non-protocol markers remain recoverable by fresh observation
  ✔ absent (1.962084ms)
  ✔ false (1.596375ms)
  ✔ string true (1.283959ms)
  ✔ nested payload (1.344209ms)
  ✔ text-only JSON (1.286167ms)
✔ ordinary pre-write failures and non-protocol markers remain recoverable by fresh observation (7.851417ms)
▶ read-only metadata and trusted not-started outcomes do not invent mutation ambiguity
  ✔ ordinary read (2.010792ms)
  ✔ bound read despite misleading write annotation (0.923875ms)
  ✔ not_started overrides a contradictory result marker (0.658917ms)
✔ read-only metadata and trusted not-started outcomes do not invent mutation ambiguity (3.747375ms)
▶ failure after an explicit unknown result cannot erase the persisted reconciliation requirement
  ✔ after_tool persistence (1.128625ms)
  ✔ tool_result callback (0.973667ms)
✔ failure after an explicit unknown result cannot erase the persisted reconciliation requirement (2.186125ms)
✔ explicit initialization and popup following complete a real workflow without executing remaining origin calls (1295.512125ms)
✔ cancellation during terminal persistence cannot return succeeded with accepted evidence (5.493792ms)
✔ deadline expiry during terminal persistence is reported and returned elapsed time includes the wait (154.496167ms)
✔ returned checkpoint accounts for final persistence time before a later resume (43.773125ms)
✔ resume retains application restrictions unless the caller explicitly replaces them (2.370417ms)
✔ resume cannot silently remove a previously required application completion policy (2.357666ms)
✔ decision checkpoint resume preserves counters and IDs but requires fresh verification (17.060334ms)
✔ an interrupted maybe-applied mutation blocks every resumed planner until trusted reconciliation (2.857834ms)
✔ failed checkpoint persistence prevents dispatch and never retries a tool (2.122542ms)
✔ restored sessions reject old session IDs and refs until a fresh snapshot is obtained (2.762042ms)
✔ untrusted checkpoint shape, protocol pairing, IDs, and budgets are validated before execution (2.323708ms)
✔ stall detection uses repeated actual outcomes, provides one replan, then stops (2.195833ms)
✔ bounded retry and fallback only repeat planning; tools with uncertain effects are not retried (3.891167ms)
✔ opt-in compaction preserves complete tool groups and real verification evidence (8.583083ms)
✔ HTTP adapter reports only actual provider usage and exposes transient status classification (10.950125ms)
✔ separate CLI processes restore private storage and fresh sessions without replaying completed writes (1871.523667ms)
✔ interrupted external side effect requires explicit reconciliation before CLI restores or plans (1750.03625ms)
✔ an exhausted saved CLI budget performs no page restoration or model request (1229.048042ms)
✔ explicit initialization traverses actual MCP before planning, with step-zero write-ahead and real evidence (13.717208ms)
✔ start URL is explicit and invalid targets fail before catalog, tools, or planning (0.613375ms)
✔ initialization obeys tool, history, deadline, and cancellation budgets without planning (32.337417ms)
✔ unknown initial effects require reconciliation and the initializer is never replayed after it (1.183041ms)
✔ an initializer that was never registered can start once when its saved prerequisite is supplied (0.4555ms)
✔ settled initialization survives resume and compaction with unique subsequent IDs (1.732209ms)
✔ version-one checkpoints migrate explicitly and initialization manifests reject inconsistent history (0.867708ms)
✔ write-ahead persistence failure never dispatches or replays the registered initializer (0.425375ms)
✔ replan_required skips later tool calls and still requires separate verification (0.325084ms)
✔ CLI start URL reaches real Chrome before its first model request and resumes without another initializer (1941.846334ms)
✔ CLI accepts a browser-wide popup policy but restricts start URL to run and validates both (1154.510292ms)
✔ agent drives real MCP browser tools and requires verified form state before finishing (710.201375ms)
✔ MCP schema rejects malformed arguments before side effects, then planner can recover (2.63925ms)
✔ stale-reference result stops the batch and is returned for replanning without retry (1.41625ms)
✔ failed checks and isError=true cannot satisfy completion even when passed=true (2.041208ms)
✔ completion rejects stale evidence, another session, invented IDs, and contradicted checks (5.399125ms)
✔ application task-specific acceptance can reject a technically passing unrelated check (0.967708ms)
✔ tool and planning budgets are bounded; invalid decisions never dispatch a tool (1.293292ms)
✔ deadline bounds a non-cooperating planner and abort propagates into an in-flight tool (23.381959ms)
✔ MCP cancellation reaches the actual server request signal (1.085666ms)
✔ HTTP planner preserves MCP image evidence and structured errors without real model calls (10.669916ms)
✔ HTTP planner rejects malformed model arguments, mixed finish calls, and oversized output (0.679ms)
✔ an installed completion policy must explicitly return true (2.355709ms)
✔ history budget stops later mutations within the same planning decision (0.531166ms)
✔ cancellation before the dispatch microtask never starts a custom tool (0.239209ms)
✔ binding is frozen trusted metadata; observations and DOM edits do not consume it or hold the queue (610.22075ms)
✔ same-URL reload invalidates a document lease even if the cached navigation counter is delayed (583.385125ms)
✔ history routes and tab switch ABA invalidate bindings permanently (569.850792ms)
✔ a navigation during the asynchronous assertion cannot produce a valid stale result (568.424709ms)
✔ signal, explicit close, session close and engine disposal revoke leases without exposing browser handles (686.422167ms)
✔ cancelled acquisition releases a real late Document handle without affecting the session (582.978666ms)
✔ pre-cancelled acquisition creates no remote handle and opaque origins are refused (615.973917ms)
✔ engine disposal revokes an in-flight acquisition and releases its late handle (505.2895ms)
✔ independent CDP navigation invalidates a lease and lease cleanup leaves the external browser alive (686.248875ms)
✔ hotel form batch navigates and independent checks verify real URL and DOM (810.965583ms)
✔ snapshots obey budgets and do not include password or hidden values (393.029459ms)
✔ a newer snapshot rejects the older revision before input (176.197625ms)
✔ a replaced DOM node with an identical label is rejected (188.987792ms)
✔ changed target semantics are rejected before a click (193.432583ms)
✔ overlay blocks input and retry requires fresh observation (1272.0765ms)
✔ semantic change during actionability waiting prevents the delayed click (438.322417ms)
✔ a base URL change invalidates the observed resolved link destination (231.310667ms)
✔ value verification never leaks a password substituted after a field-type read (272.931667ms)
✔ batch preserves completed work and reports failed and skipped steps (222.487917ms)
✔ incremental snapshots expose added and removed controls (315.569916ms)
✔ open shadow controls and explicitly selected frame controls execute (352.822875ms)
✔ sessions isolate storage and reject cross-session references (399.312584ms)
✔ concurrent operations serialize within one session (233.726ms)
✔ capture contains a real JPEG and closing removes a session (209.671291ms)
✔ CDP disposal closes only owned pages and preserves the independent browser and existing state (906.532042ms)
✔ stream and item error diagnostics may precede a successful completed turn (71.545958ms)
✔ only allowlisted structured diagnostic fields survive without deriving a retry policy (26.572667ms)
✔ a failed turn stays failed even if termination flushes completion usage and exits zero (27.487417ms)
✔ valid output and exit zero cannot substitute for an actual completed event (27.644542ms)
✔ a message-only error followed by a hanging process ends at the original deadline (259.907125ms)
✔ a nonzero exit and a completed response with invalid schema both remain failures (85.576875ms)
✔ gateway counts one successful dispatch after diagnostics and retains unknown usage on incomplete failure (112.266375ms)
✔ run failure classification uses terminal AgentFailure and never erases a returned failure (0.19075ms)
✔ Codex process decisions complete the Agent write/verify/evidence loop with isolated transient files (104.706625ms)
✔ Codex only inherits allowlisted environment and uses bounded deduplicated inline images (27.162125ms)
✔ Codex generic notifications do not abort successful terminal events or leak provider text (26.791292ms)
✔ Codex incomplete cannot dispatch or imply Agent success (26.443125ms)
✔ Codex nonzero cannot dispatch or imply Agent success (25.194666ms)
✔ Codex failed-terminal cannot dispatch or imply Agent success (24.498584ms)
✔ Codex bad-event cannot dispatch or imply Agent success (24.554125ms)
✔ Codex external cannot dispatch or imply Agent success (26.24375ms)
✔ Codex stderr-limit cannot dispatch or imply Agent success (29.44475ms)
✔ Codex large cannot dispatch or imply Agent success (27.154375ms)
✔ Codex bad-json cannot dispatch or imply Agent success (25.631792ms)
✔ Codex symlink cannot dispatch or imply Agent success (25.413917ms)
✔ Codex validates original tool arguments and the final schema before dispatch (148.206333ms)
✔ Codex timeout drains final reported usage, deletes files and keeps fixed failure diagnostics (512.737541ms)
✔ Codex one-request cancellation is isolated and close drains every outstanding process (69.668917ms)
✔ Codex missing usage counters remain absent and usage hook errors still produce process diagnostics (36.951375ms)
✔ Codex file cleanup failure remains visible to Agent and close without dropping measured usage (34.689209ms)
✔ Codex invalid configuration fails before child launch (0.13475ms)
✔ successful Codex leader cannot leave an owned stdio-independent descendant running (67.914667ms)
✔ cancelled Codex process retains usage emitted while handling SIGTERM (540.79875ms)
✔ gateway close waits for cancellation usage before exposing final metrics (92.789583ms)
✔ named tool_choice rejects a different schema-valid tool (44.193ms)
✔ tool_choice none permits a text response with tools present (30.148708ms)
✔ tool_choice auto with no selected tools is a stopped text completion (31.591417ms)
✔ timeout terminates a SIGTERM-resistant descendant after its parent exits (1712.352625ms)
✔ Codex transport adapts function decisions, disables external tools, and preserves actual CLI usage (43.683ms)
✔ Browser Use structured schema and inline images share the same Codex inference transport (30.722375ms)
✔ Codex transport rejects schema-invalid function arguments before the browser runtime sees them (31.69375ms)
✔ automatic tool choice can return ordinary text without inventing an executed call (29.227875ms)
✔ unexpected Codex tool activity invalidates the attempt and stops its own process (27.86975ms)
✔ Codex inference deadline and cancellation stop owned child processes without treating them as model answers (207.565625ms)
✔ shared model gateway bridges both framework formats through the same fake Codex process settings (90.654459ms)
✔ network-receipt comparison task requires the authenticated response and one independently judged write (1483.953666ms)
✔ Browser Use retains done, true counts, portable history and judge phase when post-task judge times out (217.476208ms)
✔ explicit judge false is preserved and ordinary full return stays distinct from cleanup (70.512958ms)
✔ judge usage contains only reported public response values, and judge exceptions retain their original phase (175.560458ms)
✔ pre-done timeout, post-done exception and non-judge finalization timeout preserve distinct observed phases (502.342041ms)
✔ owned cleanup has a bounded outcome and cannot silently count as a completed run (106.905375ms)
✔ Browser Use retains observed download paths when public kill resets browser metadata (59.839958ms)
✔ runner reports independent business success, observed agent done, and full return as separate axes after adapter failure (550.960833ms)
✔ summary retains legacy completedAndPassed while adding explicit axes (0.079ms)
✔ Tablaze flags reach the official SDK, and proposed finish stays distinct from accepted completion (168.163125ms)
✔ comparison runner stops the owned adapter process group after output overflow (646.454208ms)
✔ Browser Use source pin accepts only exact git commits or hashed official archives (57.918792ms)
✔ fixture judges actual records, rejects claimed success, and resets isolated attempts (20.039958ms)
✔ download judge requires the actual matching file, not a download request or a claimed path (3.99425ms)
✔ authorization-return judge requires one provider approval and one app submission (4.102625ms)
✔ model gateway enforces equal wire settings and measures returned usage without a paid provider (4.824583ms)
✔ missing model/dependencies stay not_run and never become a competitor failure or a win (32.735709ms)
✔ scripted Tablaze smoke completes diverse real-browser tasks with independent judges (16620.649917ms)
✔ typed backend write, actual browser receipt, and Agent verification complete one reservation (674.409792ms)
✔ real navigation while planning removes the restricted tool before its stale decision dispatches (588.021042ms)
✔ handler rechecks a real browser binding after waiting and rejects a stale external write (537.968334ms)
✔ same-origin tenant change cannot resume an uncertain write; matching identity reconciles without replay (1831.635042ms)
✔ CLI rejects a bound checkpoint before launching a browser or restoring its URLs (201.632375ms)
✔ input validation rejects extra fields and attempted context injection before entering a handler (3.702125ms)
✔ global tools work without a browser and receive trusted context, stable callId and a signal (0.98075ms)
✔ caller-selected output may deliberately include a context value (0.593375ms)
✔ published output schema validates the actual MCP envelope and relocated reusable refs (15.990375ms)
✔ write handler exceptions and invalid outputs remain unknown without exposing raw errors (2.60375ms)
✔ trusted read effects remain reads when handlers report failures (0.343125ms)
✔ duplicate registrations, forged handles and reserved names are rejected (0.454042ms)
✔ origins reject wildcard and non-origin restrictions and require a browser binding (0.260583ms)
✔ site filtering includes the exact port and metadata still covers invisible tools (6.378041ms)
✔ closed catalogs release a captured guard once and cannot dispatch handlers (0.292625ms)
✔ catalog context keys stay stable across leases and change when the captured document changes (0.506541ms)
✔ binding changes after catalog preparation refuse dispatch before entering a handler (0.734041ms)
✔ context identity is rechecked after asynchronous validation before the handler starts (0.382041ms)
✔ handler assertCurrent rejects a binding change across await and a started write remains unknown (0.410917ms)
✔ handler assertCurrent also rejects its cancelled dispatch after a noncooperating await (0.356292ms)
✔ a retained execution context is revoked when a successful dispatch has returned (0.484584ms)
✔ independent dispatch cancellation bounds a hanging getContext check before_handler (0.603584ms)
✔ independent dispatch cancellation bounds a hanging guard check before_handler (0.326792ms)
✔ independent dispatch cancellation bounds a hanging getContext check after_handler (0.4345ms)
✔ independent dispatch cancellation bounds a hanging guard check after_handler (0.390334ms)
✔ stable contract hashes exclude opaque context values while including schemas and all registry tools (0.3395ms)
✔ an already established registry rejects base-contract and tenant identity changes (0.176083ms)
✔ context resolver failures expose only a fixed registry error (0.219833ms)
✔ runAgent uses catalog dispatch, preserves callId, and records unknown custom writes for reconciliation (2.527458ms)
✔ resume rejects schema, version, effect, origin, base-contract and same-origin tenant changes before planning (6.818625ms)
✔ reconciliation of an unknown write cannot rebind its checkpoint to another tenant on the same origin (0.989167ms)
✔ runAgent discards a decision when the captured browser binding changes during planning (0.505417ms)
✔ viewport observation reaches the 501st button after scrolling a truncated page (726.083625ms)
✔ scoped observation exposes the selected subtree and resets incompatible diff baselines (199.788417ms)
✔ viewport text clips individual offscreen lines within a single text node (171.141292ms)
✔ viewport observation excludes controls and text clipped by scroll-container ancestry (164.670292ms)
✔ snapshots, extraction, verification and guarded actions share open shadow and slot content (228.395334ms)
✔ hidden hosts, hidden slots and raw form values do not leak through composed readers (281.133417ms)
✔ text matching searches beyond bounded evidence and across composed text-node boundaries (194.570333ms)
✔ explicit hidden accessibility labels still name guarded controls without leaking form values (226.048625ms)
✔ link and table extraction report item and character truncation accurately (197.14675ms)
✔ scan-budget exhaustion is explicit in both snapshots and extraction (280.761375ms)
✔ ordinary draggable and aria-labeled divs deliver actual native drop payload (735.284958ms)
✔ mouse-driven custom drag supports the observed source covering its drop target (864.531375ms)
✔ nested scaled iframe drag maps main-page coordinates and preserves native payload (1426.253167ms)
✔ open Shadow DOM drag target hit testing reaches the actual source and destination (736.969458ms)
✔ an overlay appearing after the initial mouse move cannot intercept mousedown (1272.839666ms)
✔ an overlay appearing during movement cancels the drag before drop and releases the mouse (931.8695ms)
✔ mouseup RPC rejection is a failed action even if the release was applied; cleanup releases held input (1378.680084ms)
✔ an unreleasable mouse closes only the owned session and reports input cleanup failure (892.288166ms)
✔ schema extraction binds every populated leaf to exact source text and authoritative URLs (15.296292ms)
✔ multiple sources can produce nested data and a checked aggregate without a provider dependency (6.434417ms)
✔ invented sources, URLs, approximate quotes and missing or non-leaf citations are rejected (22.967125ms)
✔ JSON pointers escape literal keys and require evidence for null and empty containers (2.238084ms)
✔ schema types, required fields, formats and local references are enforced without coercion (10.253166ms)
✔ literal schema const data is not interpreted as schema references (3.613167ms)
✔ source and candidate limits fail before accepting output (14.103959ms)
✔ quote presence is distinct from truth and the application can reject unsupported claims (17.358042ms)
✔ cancellation and deadline stop waiting on non-cooperating extractors without retries (16.965333ms)
✔ deterministic fields create typed values and preserve their exact DOM provenance (1.866041ms)
✔ field-map cardinality, optional omissions, strict conversions and observation binding are enforced (20.376334ms)
✔ browser inspector uses composed text, current form state and real page provenance (638.808917ms)
✔ find traverses actual virtualized rows and returns a fresh actionable viewport ref (4881.760666ms)
✔ MCP find supports a selected iframe and preserves current snapshot semantics (795.564708ms)
✔ closing a CDP session cancels its unfinished download without closing the external browser (777.14575ms)
✔ a tab created after disposal begins is closed instead of escaping session ownership (757.457959ms)
✔ coordinate clicks never move to a sibling tab when the observed tab closes (804.286ms)
✔ allowed native form submission preserves one POST with the actual encoded business values (758.452041ms)
✔ allowed 307 redirect preserves the original POST method and body without duplicate dispatch (716.153083ms)
✔ same-origin 302 processes Set-Cookie and a matching workspace restores that cookie and policy (1056.651084ms)
✔ explicit open/goto/new-tab deny the other port before a request or a new tab exists (752.968791ms)
✔ multi-hop HTTP 302 is intercepted before the denied hop during open, goto and new-tab (921.375667ms)
✔ iframe redirect is blocked before its denied server receives a request and counts remain per context (1004.887625ms)
✔ one context blocking a document does not interrupt a different context pending business action (1217.411458ms)
✔ a new popup first-document redirect is blocked even before an ordinary popup snapshot exists (854.802708ms)
✔ page script and meta refresh cannot issue an HTTP document request to a denied origin (1061.517541ms)
✔ an allowed localhost child of the 127.0.0.1 page retains interception through its cross-site redirect (1010.614125ms)
✔ non-document fetch and image requests reach the other origin: navigation policy is not a network firewall (483.338125ms)
✔ workspace policy changes, missing hashes, policy removal and newly added policy reject before browser launch (0.74325ms)
✔ CDP with configured navigation policy is rejected by the constructor without a connection attempt (52.526625ms)
✔ absent navigation policy retains ordinary browsing and does not attach a policy hash or diagnostics (611.602916ms)
✔ doctor validates and summarizes canonical policy without launching or printing origins (777.195042ms)
✔ policy file parsing rejects unsafe shapes, origin syntax, unknown fields and raw error disclosure (2902.311667ms)
✔ policy reader enforces a 64 KiB file bound and refuses directories and missing files (774.232334ms)
✔ policy reader refuses a FIFO without waiting for a writer (197.660625ms)
✔ policy and external CDP or setup are rejected before reading the policy path (768.758833ms)
✔ malformed private checkpoints do not disclose JSON excerpts before policy restoration (583.43775ms)
✔ MCP startup accepts a valid policy and exposes tools without launching a browser (207.813416ms)
✔ empty CLI checkpoints retain policy identity and reject changed, removed or legacy policy on resume (1236.60275ms)
✔ dispose closes its already-owned browser while policy bootstrap is still awaiting a CDP response (2.456958ms)
✔ an in-flight real navigation reports a safe policy fault instead of the transport URL/error (651.845917ms)
✔ policy failure after a real write retains partial action evidence and skips the next mutation (706.047958ms)
✔ undefined disables policy while configured empty policy permits only HTTP(S) (0.655875ms)
✔ explicit empty allowlist denies all targets and differs from an absent allowlist (0.309125ms)
✔ exact origin matching preserves scheme, port and host boundaries with blocked precedence (0.225042ms)
✔ WHATWG normalization covers case, default ports, Unicode hostnames, IPv4 and IPv6 (0.10675ms)
✔ canonical identity is versioned, stable across equivalent ordering and does not collapse deny-all (0.20975ms)
✔ compiled policy and copied arrays cannot be mutated and detached methods retain their policy (0.210292ms)
✔ only origins are accepted as rules; URL parser repairs and wildcard guesses are rejected (0.840333ms)
✔ strict bounded configuration rejects unknown fields, wrong types, sparse arrays and accessors (0.824375ms)
✔ combined allow/block input is limited to 200 entries before normalization or deduplication (0.63375ms)
✔ invalid configuration, malformed navigation and denied URLs never surface private input (0.399542ms)
✔ opt-in MCP journal correlates an authenticated popup response without exposing request query or headers (911.1915ms)
✔ network capture is absent by default and refuses bodies without a declared bound (593.459375ms)
✔ attached CDP capture ignores independent Chrome pages (625.590709ms)
✔ response bodies remain gated after a scoped credential is filled (614.459791ms)
✔ pre-cancelled open does not start a browser or create a session (3.400291ms)
✔ cancellation during shared browser startup returns promptly and leaves it available (598.341875ms)
✔ a context that arrives after cancellation is closed without touching a sibling session (541.210792ms)
✔ page creation cancellation closes the owned context and any late page (715.287459ms)
✔ navigation cancellation removes only the opening attempt and drains its resources (717.626125ms)
✔ snapshot cancellation cannot register a completed late observation as a live session (660.602ms)
✔ MCP tab_open propagates cancellation to the engine while sibling sessions survive (700.242583ms)
✔ CDP cancellation cleans a late owned page and leaves independent pages and browser alive (1037.365083ms)
✔ dispose is bounded while an isolated context acquisition has not returned (2455.345833ms)
✔ dispose is bounded while an isolated page acquisition has not returned (2524.582958ms)
✔ dispose reports context close failures rather than claiming complete cleanup (638.24475ms)
✔ dispose retains an existing session cleanup after close removes the session from its map (2340.625959ms)
✔ CDP disposal reports an unidentified pending page and closes it after late arrival (2793.807458ms)
✔ planner removes only exact plain duplicates while preserving errors and distinct or annotated text (10.495458ms)
✔ planner retains text-only fallback and near-duplicate serialization (0.264542ms)
✔ projection preserves image order and complete assistant/tool groups with and without image support (0.55525ms)
✔ projection leaves successful evidence, full history, events, and saved checkpoints intact (3.65175ms)
✔ default stay keeps the opener active and completes remaining actions (730.747875ms)
✔ follow-single returns popup observation and truthfully skips remaining old-ref actions (891.433292ms)
✔ multiple candidates are retained without choosing the first or stealing focus (832.821917ms)
✔ a late popup is owned but never retroactively changes the active tab (1549.482959ms)
✔ a different owned background opener cannot enter the foreground action window (944.369417ms)
✔ iframe-triggered unique popup is associated with the observed owning page (929.931417ms)
✔ about:blank popups return bounded observations without waiting for eventual navigation (902.635708ms)
✔ popup collection consumes the existing batch deadline and never retries the input (676.188709ms)
✔ workspace preserves policy and old version-one records restore the stay default (2046.969ms)
✔ failed workspace restoration rolls back policy as well as newly owned sessions (538.449667ms)
✔ stdio handshake exposes the bounded browser tools and structured session errors (212.003959ms)
✔ stdio tools drive isolated Chromium and return browser evidence (955.202708ms)
✔ SDK cancellation stops the browser batch before its later side effect (872.692041ms)
✔ CDP connection errors never return endpoint credentials (218.978667ms)
✔ a total batch budget ends a longer individual wait and reports skipped work (992.973292ms)
✔ CLI help, version and doctor are usable without launching a browser (772.415208ms)
✔ provider selection preserves required explicit model and rejects mismatched options before execution (2894.488084ms)
✔ new provider-only options cannot start the default MCP server (762.43825ms)
✔ Anthropic CLI forwards its explicit model, native tool schema, key and output cap; retains real usage (214.910833ms)
✔ Ollama CLI uses native requests, optional num_predict and no invented token counters (420.970917ms)
✔ explicit compatible provider keeps the previous endpoint and key contract (212.408166ms)
✔ Codex CLI uses the chosen executable and explicit model/effort without a model endpoint (666.033084ms)
✔ Codex executable defaults to codex on PATH without inventing a reasoning setting (666.435458ms)
✔ Codex failures expose fixed diagnostics without stderr, credentials or provider error text (649.361084ms)
✔ Codex provider integrates start-url, actual browser verification and normal cleanup (1293.986708ms)
✔ Codex checkpoint resume preserves prior budgets and still requires provider/model on each invocation (902.575334ms)
✔ bound checkpoints are rejected before any Codex process or browser restoration (199.081125ms)
✔ SIGTERM cancels an in-flight Codex planner and terminates its child (638.949458ms)
✔ anthropic: native wire request completes a verified real MCP tool loop (32.827ms)
✔ anthropic: images and errors preserve paired repeated-name tool results without mutating history (4.535041ms)
✔ anthropic: text-only selection omits image bytes and native defaults remain explicit (1.544625ms)
✔ anthropic: malformed native decisions, mixed controls and unknown control fields fail safely (5.812042ms)
✔ anthropic: HTTP diagnostics discard private error bodies and do not retry (2.011125ms)
✔ anthropic: native response size is bounded before dropping unknown fields (1.10825ms)
✔ anthropic: cancellation interrupts a pending native response body (1.504792ms)
✔ anthropic: actual missing or invalid usage counters stay absent (1.112417ms)
✔ anthropic: malformed JSON and interrupted bodies retain distinct safe response failures (53.997ms)
✔ anthropic: unpaired history is rejected before any native request (0.5315ms)
✔ ollama: native wire request completes a verified real MCP tool loop (4.995208ms)
✔ ollama: images and errors preserve paired repeated-name tool results without mutating history (0.965459ms)
✔ ollama: text-only selection omits image bytes and native defaults remain explicit (1.557083ms)
✔ ollama: malformed native decisions, mixed controls and unknown control fields fail safely (3.782291ms)
✔ ollama: HTTP diagnostics discard private error bodies and do not retry (1.058ms)
✔ ollama: native response size is bounded before dropping unknown fields (0.884542ms)
✔ ollama: cancellation interrupts a pending native response body (0.910167ms)
✔ ollama: actual missing or invalid usage counters stay absent (0.75925ms)
✔ ollama: malformed JSON and interrupted bodies retain distinct safe response failures (54.50025ms)
✔ ollama: unpaired history is rejected before any native request (0.259208ms)
✔ native adapters validate configuration and use only their configured/default native endpoint (0.552792ms)
✔ native factory header validation never includes private names, values or keys in exceptions (0.530958ms)
✔ spawned native-provider CLI rejects an invalid environment key without echoing it to stderr (400.94375ms)
✔ real MCP form fill/save verifies current refs and visible status without another snapshot (750.9215ms)
✔ read-only ref checks preserve actionability and can read a consumed current revision (254.730084ms)
✔ ref checks support textarea, select, empty values and open shadow controls (284.400583ms)
✔ MCP and engine reject missing/ambiguous targets and a ref without snapshot_id (189.435375ms)
✔ new observations and other sessions reject old snapshot identities; unknown refs fail (352.810708ms)
✔ replaced, renamed, hidden and navigated targets are rejected without selector fallback (961.532375ms)
✔ parallel status waits cannot preserve ref evidence from a replaced node or old document (823.847458ms)
✔ a later status cannot keep a previously matching value after the same node changes (421.493416ms)
✔ iframe refs retain their frame identity and reject detached frame evidence (388.593292ms)
✔ sensitive and non-form targets fail without leaking values; evidence remains bounded (351.120333ms)
✔ queued verification waits for the action batch and shares the original readonly revision (435.304541ms)
✔ scoped credential completes a real login and never appears in normal evidence or artifacts (1031.333417ms)
✔ iframe credentials require both the frame and top-level origin to match (807.472292ms)
✔ MCP uses only the credential alias and reports login evidence without disclosing the value (2731.250292ms)
✔ doctor reports only secret metadata and never launches the configured browser (402.216208ms)
✔ malformed secret configuration fails before doctor or run with one fixed message (1162.584375ms)
✔ CDP and setup reject secrets before reading the configuration path (773.06625ms)
✔ stdio accepts configured aliases and exposes the public action without resolving or launching (209.628667ms)
✔ CLI run saves secret policy identity and requires the same configuration on resume (831.027958ms)
✔ configuration does not read env values until an allowed-origin resolution and supports rotation (4.970959ms)
✔ returned configuration and scopes are frozen, with only explicit artifact opt-in (1.366ms)
✔ resolvers reject missing, inherited, empty and oversized values without disclosing their source (1.2575ms)
✔ aborted resolution does not read env and an abort during a getter cannot return the value (0.596834ms)
✔ strict JSON shape and shared SDK metadata limits reject malformed private configurations (3.999625ms)
✔ reader enforces the actual 64 KiB boundary, UTF-8 and regular files with fixed errors (0.906167ms)
✔ reader refuses a FIFO without blocking on a writer (2.866125ms)
✔ pre-page bridge exposes only bounded metadata and fills the original control without focusing a decoy (458.138958ms)
✔ old handles fail after identical DOM replacement without filling the replacement (171.046084ms)
✔ opaque HTTP sandbox remains denied despite replacing window.origin in page script (184.25625ms)
✔ captured native primitives withstand later prototype and global replacements (147.479167ms)
✔ bridge property and operation references cannot be replaced by page code (149.510667ms)
✔ invalid bindings, values and unsupported or disabled controls reject before writing (165.079584ms)
✔ an input listener detaching the filled node reports wrote=true and does not dispatch change or fill its replacement (154.564833ms)
✔ same-URL navigation generates a fresh document identity and rejects the prior binding (162.03925ms)
✔ text projection uses lookahead before clipping and never promotes its trailing content (1.073916ms)
✔ raw boundaries remain exact when a secret resembles the replacement marker or overlaps another secret (0.858916ms)
✔ snapshot projection covers fields, options, diffs and metadata without changing reference identity (0.54725ms)
✔ link extraction preserves href-first raw allocation when redaction shrinks the last field (0.15175ms)
✔ link text and table cells redact before their field and aggregate boundaries (0.134708ms)
✔ short secrets cannot expand links or table cells past the aggregate public budget (1.727125ms)
✔ expansion followed by shrinking never reallocates the original reader budget to later fields (0.572708ms)
✔ text extraction restores the original 20000 character boundary (0.09225ms)
✔ selected metadata is copied and recursively redacted with bounded rejection (0.282458ms)
✔ metadata bounds replacement expansion as well as raw input size (45.215875ms)
✔ published and runtime MCP schemas accept only a bounded alias plus observed ref for fill_secret (29.629208ms)
✔ secret failures retain their fixed public code in structured and text MCP results (1.340834ms)
✔ SDK server rejects secret configuration with external CDP before any resolver is called (0.223708ms)
✔ undefined disables secrets; compiled facade is frozen, detached methods work, and values are private (1.707667ms)
✔ strict options reject unknown keys, accessors, prototypes and wrong optional types without invoking getters (0.351958ms)
✔ strict definitions reject invalid names, versions, duplicate aliases, unknown fields and inherited resolvers (0.490042ms)
✔ arrays reject holes, extra properties and getters; alias and configured-origin limits are bounded (1.073459ms)
✔ origin lists require exact nonempty HTTP(S) origins; wildcard, credentials, path and opaque origins are rejected (0.267ms)
✔ frame and top origins are both required, including port and scheme; scope failure never invokes resolver (0.422167ms)
✔ hash normalizes ordering, duplicates and explicit default top scope; excludes callbacks and resolved values (0.528208ms)
✔ caller mutations cannot change cloned scopes, metadata or captured resolver (0.134167ms)
✔ context is checked before resolving and again before delivering; failed delivery still redacts the resolved value (0.243166ms)
✔ resolver and context exceptions are fixed and never expose raw thrown objects or messages (0.607625ms)
✔ pre-aborted calls invoke neither context nor resolver and never expose the abort reason (0.132167ms)
✔ cancellation bounds a hanging pre-context check and does not enter the resolver later (2.278917ms)
✔ cancellation bounds resolver wait, passes an aborted signal, and registers its late value while store remains alive (0.359958ms)
✔ late resolver rejection after cancellation is observed without unhandled rejection (0.226167ms)
✔ cancellation bounds a hanging post-resolution context check while preserving redaction (0.215791ms)
✔ all rotated and cross-alias values remain redacted; duplicate values do not consume retention slots (1.832208ms)
✔ resolved values must be nonempty bounded well-formed strings and errors contain no returned value (0.28825ms)
✔ 128 distinct-value limit fails before delivery without dropping earlier redactions (2.126417ms)
✔ 64 KiB retention limit counts UTF-8 bytes and rejects overflow atomically (10.084209ms)
✔ raw, whitespace-normalized, URI and JSON-escaped variants redact literally and contains follows the same set (0.307833ms)
✔ overlapping secrets use longest simultaneous replacement and never reprocess the replacement marker (0.169208ms)
✔ crossing and chained matches cover every sensitive character in original coordinates (0.1885ms)
✔ public aliases equal to known values are hidden without changing configured resolution or unrelated names (0.155833ms)
✔ a late cancelled value and a newer rotated value are both retained until close (1.396333ms)
✔ padding covers maximum URI and JSON forms; redaction before truncation does not reveal a secret prefix (2.763458ms)
✔ close revokes active resolution, aborts the resolver signal and prevents silent redaction bypass afterward (0.329333ms)
✔ MCP schema extraction returns typed shadow/form data and keeps observed references usable (569.609042ms)
✔ MCP preserves structured validation error codes and pointer issues without raw page values (199.0065ms)
✔ MCP rejects sensitive, hidden, raw-form and truncated fields without partial extracted data (218.644959ms)
✔ MCP extraction follows the explicitly observed frame and cites its own URL (224.072709ms)
✔ MCP bounds field counts and rejects unsupported field options before extraction (202.532417ms)
✔ real actions show a non-intercepting pointer on owned pages and rejected refs do not fake a click (3430.885291ms)
✔ operators can disable the visual pointer (630.170083ms)
✔ sessions do not add a pointer by default, even when headed (744.155916ms)
✔ popup tabs stay owned, switching invalidates references and last close ends session (567.237166ms)
✔ navigation and explicit tabs preserve session state and handle browser history (364.691334ms)
✔ upload reads an explicit local fixture and download produces the actual file (316.827917ms)
✔ native confirm and prompt use one-shot policies and disclose actual results (436.745375ms)
✔ storage state restores cookies and localStorage across independent engine lifetimes (1020.772041ms)
✔ hover and double-click execute guarded references, coordinates reject offscreen input (322.72625ms)
✔ CSS-pixel screenshots support actual canvas coordinate input (237.220125ms)
ℹ tests 443
ℹ suites 0
ℹ pass 443
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 184796.690834
