
> 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 (4394.078208ms)
✔ PDF metadata stays bound to its captured tab through asynchronous artifact persistence (609.741375ms)
✔ drag revalidates observed targets after asynchronous geometry reads before mouse input (664.872709ms)
✔ upload chooser revalidates its visible trigger after asynchronous local file checks (532.811583ms)
✔ newly discovered scroll containers retain node identity guards when replaced (575.099208ms)
✔ scrollable containers expose ranges and accept guarded horizontal and vertical input (546.085458ms)
✔ dragging emits the browser drop event and delivers the real data payload (705.002833ms)
✔ hidden file chooser transfers explicit local bytes without a hidden DOM reference (718.186542ms)
✔ PDF exports real private artifacts with verifiable digest and source (551.226625ms)
✔ workspace roundtrip restores isolated durable storage, owned tabs, and active selection (2018.636917ms)
✔ workspace validation precedes navigation and failed restore rolls back owned sessions (910.805833ms)
✔ per-decision catalogs hide execution metadata and context changes skip queued writes (12.994459ms)
✔ a context change during planning or write-ahead persistence never executes the stale handler (3.346583ms)
✔ trusted write effects retain ambiguity for thrown, malformed, error, and unknown outcomes (8.208584ms)
✔ finish rechecks its lease and cannot reuse verification from the old context (3.19525ms)
✔ a context change between closed and newly prepared leases invalidates old verification (1.776292ms)
✔ only a confirmed session close preserves current evidence across context changes and later writes still invalidate it (4.708292ms)
✔ bound checkpoints reject missing runtimes and changed registry or same-origin tenant identities (1.135917ms)
✔ version-one and version-two checkpoints explicitly migrate unbound and cannot gain a registry (1.260208ms)
✔ resume rechecks a pending call effect rather than trusting a tampered mutating flag (1.229333ms)
✔ initialization uses the bound dispatcher and each lease closes before terminal persistence (1.409209ms)
✔ a prepared lease that arrives after cancellation is still closed without planning (2.707208ms)
✔ catalog release failure or cancellation cannot turn accepted verification into success (2.057333ms)
✔ run CLI completes a scripted model task through real Chrome and prints a bounded report (899.216625ms)
✔ run CLI returns human input with exit 2 and an explicit failure with exit 1 (420.688792ms)
✔ run CLI does not accept an unverified model completion and enforces the step limit (228.542625ms)
✔ CLI rejects agent-only options outside run and validates required arguments and limits (2302.411541ms)
✔ CLI reports cleanup failure separately from an already verified successful agent result (809.715542ms)
✔ restoration abort handles a rejected asynchronous dispose without an unhandled rejection (325.804375ms)
✔ CLI publishes bounded planner diagnostics without exposing endpoint, key, or upstream error text (222.977292ms)
✔ HTTP planner failures expose only status and fixed classification without enabling default retries (19.945833ms)
✔ unknown exceptions cannot inject diagnostic codes and custom retry policy remains authoritative (2.820083ms)
✔ transport and malformed response diagnostics remain distinct and omit raw response data (3.387084ms)
✔ catalog failure is distinguishable before any planner or mutation runs (1.787792ms)
✔ application callbacks are classified without recursively notifying a broken event hook (3.135583ms)
✔ adapter usage callback failures retain their application phase and original retry policy input (1.032333ms)
✔ persistence failures prevent mutation or successful completion without changing resume schema (4.408625ms)
✔ a noncooperating checkpoint callback receives a bounded persistence-timeout diagnosis (5004.585875ms)
✔ ordinary MCP explicit unknown write persists uncertainty, skips the batch and cannot replay on resume (16.980291ms)
▶ explicit uncertainty overrides a superficially successful write in legacy and bound MCP execution
  ✔ ordinary MCP write annotation (1.560041ms)
  ✔ bound trusted write metadata (3.528875ms)
✔ explicit uncertainty overrides a superficially successful write in legacy and bound MCP execution (5.270792ms)
▶ ordinary pre-write failures and non-protocol markers remain recoverable by fresh observation
  ✔ absent (1.862542ms)
  ✔ false (1.804334ms)
  ✔ string true (1.345084ms)
  ✔ nested payload (1.35675ms)
  ✔ text-only JSON (1.3255ms)
✔ ordinary pre-write failures and non-protocol markers remain recoverable by fresh observation (8.001334ms)
▶ read-only metadata and trusted not-started outcomes do not invent mutation ambiguity
  ✔ ordinary read (2.053666ms)
  ✔ bound read despite misleading write annotation (0.886125ms)
  ✔ not_started overrides a contradictory result marker (0.661375ms)
✔ read-only metadata and trusted not-started outcomes do not invent mutation ambiguity (3.7565ms)
▶ failure after an explicit unknown result cannot erase the persisted reconciliation requirement
  ✔ after_tool persistence (1.316416ms)
  ✔ tool_result callback (1.069875ms)
✔ failure after an explicit unknown result cannot erase the persisted reconciliation requirement (2.48325ms)
✔ explicit initialization and popup following complete a real workflow without executing remaining origin calls (1276.608541ms)
✔ cancellation during terminal persistence cannot return succeeded with accepted evidence (5.393666ms)
✔ deadline expiry during terminal persistence is reported and returned elapsed time includes the wait (154.346042ms)
✔ returned checkpoint accounts for final persistence time before a later resume (42.745ms)
✔ resume retains application restrictions unless the caller explicitly replaces them (2.106708ms)
✔ resume cannot silently remove a previously required application completion policy (2.286625ms)
✔ decision checkpoint resume preserves counters and IDs but requires fresh verification (16.553917ms)
✔ an interrupted maybe-applied mutation blocks every resumed planner until trusted reconciliation (2.845417ms)
✔ failed checkpoint persistence prevents dispatch and never retries a tool (2.393083ms)
✔ restored sessions reject old session IDs and refs until a fresh snapshot is obtained (2.289792ms)
✔ untrusted checkpoint shape, protocol pairing, IDs, and budgets are validated before execution (2.14275ms)
✔ stall detection uses repeated actual outcomes, provides one replan, then stops (2.105167ms)
✔ bounded retry and fallback only repeat planning; tools with uncertain effects are not retried (4.333833ms)
✔ opt-in compaction preserves complete tool groups and real verification evidence (7.769416ms)
✔ HTTP adapter reports only actual provider usage and exposes transient status classification (11.182125ms)
✔ separate CLI processes restore private storage and fresh sessions without replaying completed writes (1847.172709ms)
✔ interrupted external side effect requires explicit reconciliation before CLI restores or plans (1759.147875ms)
✔ an exhausted saved CLI budget performs no page restoration or model request (1202.241666ms)
✔ explicit initialization traverses actual MCP before planning, with step-zero write-ahead and real evidence (13.037042ms)
✔ start URL is explicit and invalid targets fail before catalog, tools, or planning (0.562ms)
✔ initialization obeys tool, history, deadline, and cancellation budgets without planning (33.744917ms)
✔ unknown initial effects require reconciliation and the initializer is never replayed after it (1.018584ms)
✔ an initializer that was never registered can start once when its saved prerequisite is supplied (0.4025ms)
✔ settled initialization survives resume and compaction with unique subsequent IDs (1.769834ms)
✔ version-one checkpoints migrate explicitly and initialization manifests reject inconsistent history (0.815667ms)
✔ write-ahead persistence failure never dispatches or replays the registered initializer (0.386584ms)
✔ replan_required skips later tool calls and still requires separate verification (0.32075ms)
✔ CLI start URL reaches real Chrome before its first model request and resumes without another initializer (1941.602459ms)
✔ CLI accepts a browser-wide popup policy but restricts start URL to run and validates both (1151.70525ms)
✔ agent drives real MCP browser tools and requires verified form state before finishing (706.4295ms)
✔ MCP schema rejects malformed arguments before side effects, then planner can recover (2.58325ms)
✔ stale-reference result stops the batch and is returned for replanning without retry (1.407625ms)
✔ failed checks and isError=true cannot satisfy completion even when passed=true (1.984042ms)
✔ completion rejects stale evidence, another session, invented IDs, and contradicted checks (4.497875ms)
✔ application task-specific acceptance can reject a technically passing unrelated check (0.89825ms)
✔ tool and planning budgets are bounded; invalid decisions never dispatch a tool (1.243291ms)
✔ deadline bounds a non-cooperating planner and abort propagates into an in-flight tool (22.757292ms)
✔ MCP cancellation reaches the actual server request signal (1.018ms)
✔ HTTP planner preserves MCP image evidence and structured errors without real model calls (10.265167ms)
✔ HTTP planner rejects malformed model arguments, mixed finish calls, and oversized output (0.697542ms)
✔ an installed completion policy must explicitly return true (2.7575ms)
✔ history budget stops later mutations within the same planning decision (0.640209ms)
✔ cancellation before the dispatch microtask never starts a custom tool (0.266792ms)
✔ binding is frozen trusted metadata; observations and DOM edits do not consume it or hold the queue (586.128542ms)
✔ same-URL reload invalidates a document lease even if the cached navigation counter is delayed (607.854166ms)
✔ history routes and tab switch ABA invalidate bindings permanently (643.35525ms)
✔ a navigation during the asynchronous assertion cannot produce a valid stale result (601.521584ms)
✔ signal, explicit close, session close and engine disposal revoke leases without exposing browser handles (705.424208ms)
✔ cancelled acquisition releases a real late Document handle without affecting the session (600.610542ms)
✔ pre-cancelled acquisition creates no remote handle and opaque origins are refused (538.366292ms)
✔ engine disposal revokes an in-flight acquisition and releases its late handle (584.074625ms)
✔ independent CDP navigation invalidates a lease and lease cleanup leaves the external browser alive (702.396ms)
✔ hotel form batch navigates and independent checks verify real URL and DOM (823.40025ms)
✔ snapshots obey budgets and do not include password or hidden values (378.034ms)
✔ a newer snapshot rejects the older revision before input (184.905125ms)
✔ a replaced DOM node with an identical label is rejected (186.007458ms)
✔ changed target semantics are rejected before a click (193.262208ms)
✔ overlay blocks input and retry requires fresh observation (1287.891916ms)
✔ semantic change during actionability waiting prevents the delayed click (320.635833ms)
✔ a base URL change invalidates the observed resolved link destination (249.901459ms)
✔ value verification never leaks a password substituted after a field-type read (280.178292ms)
✔ batch preserves completed work and reports failed and skipped steps (225.580042ms)
✔ incremental snapshots expose added and removed controls (340.838875ms)
✔ open shadow controls and explicitly selected frame controls execute (343.708208ms)
✔ sessions isolate storage and reject cross-session references (433.767583ms)
✔ concurrent operations serialize within one session (221.601791ms)
✔ capture contains a real JPEG and closing removes a session (211.113584ms)
✔ CDP disposal closes only owned pages and preserves the independent browser and existing state (879.787083ms)
✔ stream and item error diagnostics may precede a successful completed turn (70.992417ms)
✔ only allowlisted structured diagnostic fields survive without deriving a retry policy (26.954333ms)
✔ a failed turn stays failed even if termination flushes completion usage and exits zero (27.929959ms)
✔ valid output and exit zero cannot substitute for an actual completed event (26.181667ms)
✔ a message-only error followed by a hanging process ends at the original deadline (260.734542ms)
✔ a nonzero exit and a completed response with invalid schema both remain failures (87.98625ms)
✔ gateway counts one successful dispatch after diagnostics and retains unknown usage on incomplete failure (115.339125ms)
✔ run failure classification uses terminal AgentFailure and never erases a returned failure (0.166375ms)
✔ Codex process decisions complete the Agent write/verify/evidence loop with isolated transient files (104.435125ms)
✔ Codex only inherits allowlisted environment and uses bounded deduplicated inline images (29.264417ms)
✔ Codex generic notifications do not abort successful terminal events or leak provider text (30.524042ms)
✔ Codex incomplete cannot dispatch or imply Agent success (26.051833ms)
✔ Codex nonzero cannot dispatch or imply Agent success (29.052458ms)
✔ Codex failed-terminal cannot dispatch or imply Agent success (24.186084ms)
✔ Codex bad-event cannot dispatch or imply Agent success (25.039792ms)
✔ Codex external cannot dispatch or imply Agent success (24.502792ms)
✔ Codex stderr-limit cannot dispatch or imply Agent success (26.758625ms)
✔ Codex large cannot dispatch or imply Agent success (28.029125ms)
✔ Codex bad-json cannot dispatch or imply Agent success (25.133375ms)
✔ Codex symlink cannot dispatch or imply Agent success (25.206584ms)
✔ Codex validates original tool arguments and the final schema before dispatch (134.340042ms)
✔ Codex timeout drains final reported usage, deletes files and keeps fixed failure diagnostics (513.5395ms)
✔ Codex one-request cancellation is isolated and close drains every outstanding process (70.592292ms)
✔ Codex missing usage counters remain absent and usage hook errors still produce process diagnostics (33.464167ms)
✔ Codex file cleanup failure remains visible to Agent and close without dropping measured usage (32.712791ms)
✔ Codex invalid configuration fails before child launch (0.154708ms)
✔ successful Codex leader cannot leave an owned stdio-independent descendant running (70.4755ms)
✔ cancelled Codex process retains usage emitted while handling SIGTERM (538.96525ms)
✔ gateway close waits for cancellation usage before exposing final metrics (97.817667ms)
✔ named tool_choice rejects a different schema-valid tool (43.412958ms)
✔ tool_choice none permits a text response with tools present (30.881042ms)
✔ tool_choice auto with no selected tools is a stopped text completion (32.234125ms)
✔ timeout terminates a SIGTERM-resistant descendant after its parent exits (1716.753584ms)
✔ Codex transport adapts function decisions, disables external tools, and preserves actual CLI usage (45.164667ms)
✔ Browser Use structured schema and inline images share the same Codex inference transport (31.645375ms)
✔ Codex transport rejects schema-invalid function arguments before the browser runtime sees them (31.918709ms)
✔ automatic tool choice can return ordinary text without inventing an executed call (28.40375ms)
✔ unexpected Codex tool activity invalidates the attempt and stops its own process (27.91825ms)
✔ Codex inference deadline and cancellation stop owned child processes without treating them as model answers (210.880416ms)
✔ shared model gateway bridges both framework formats through the same fake Codex process settings (93.746917ms)
✔ network-receipt comparison task requires the authenticated response and one independently judged write (1569.097333ms)
✔ Browser Use retains done, true counts, portable history and judge phase when post-task judge times out (219.509167ms)
✔ explicit judge false is preserved and ordinary full return stays distinct from cleanup (67.436167ms)
✔ judge usage contains only reported public response values, and judge exceptions retain their original phase (172.631458ms)
✔ pre-done timeout, post-done exception and non-judge finalization timeout preserve distinct observed phases (493.769333ms)
✔ owned cleanup has a bounded outcome and cannot silently count as a completed run (100.633583ms)
✔ Browser Use retains observed download paths when public kill resets browser metadata (60.891167ms)
✔ runner reports independent business success, observed agent done, and full return as separate axes after adapter failure (313.109ms)
✔ summary retains legacy completedAndPassed while adding explicit axes (0.07425ms)
✔ Tablaze flags reach the official SDK, and proposed finish stays distinct from accepted completion (165.167291ms)
✔ comparison runner stops the owned adapter process group after output overflow (638.248875ms)
✔ Browser Use source pin accepts only exact git commits or hashed official archives (56.914875ms)
✔ fixture judges actual records, rejects claimed success, and resets isolated attempts (20.542875ms)
✔ download judge requires the actual matching file, not a download request or a claimed path (4.097042ms)
✔ authorization-return judge requires one provider approval and one app submission (3.77675ms)
✔ model gateway enforces equal wire settings and measures returned usage without a paid provider (5.702791ms)
✔ missing model/dependencies stay not_run and never become a competitor failure or a win (34.552542ms)
✔ scripted Tablaze smoke completes diverse real-browser tasks with independent judges (16789.169792ms)
✔ typed backend write, actual browser receipt, and Agent verification complete one reservation (704.360625ms)
✔ real navigation while planning removes the restricted tool before its stale decision dispatches (601.168209ms)
✔ handler rechecks a real browser binding after waiting and rejects a stale external write (534.866208ms)
✔ same-origin tenant change cannot resume an uncertain write; matching identity reconciles without replay (1663.201042ms)
✔ CLI rejects a bound checkpoint before launching a browser or restoring its URLs (200.606ms)
✔ input validation rejects extra fields and attempted context injection before entering a handler (3.437083ms)
✔ global tools work without a browser and receive trusted context, stable callId and a signal (0.969833ms)
✔ caller-selected output may deliberately include a context value (0.592125ms)
✔ published output schema validates the actual MCP envelope and relocated reusable refs (16.666125ms)
✔ write handler exceptions and invalid outputs remain unknown without exposing raw errors (2.484959ms)
✔ trusted read effects remain reads when handlers report failures (0.31425ms)
✔ duplicate registrations, forged handles and reserved names are rejected (0.435833ms)
✔ origins reject wildcard and non-origin restrictions and require a browser binding (0.272417ms)
✔ site filtering includes the exact port and metadata still covers invisible tools (6.31525ms)
✔ closed catalogs release a captured guard once and cannot dispatch handlers (0.397375ms)
✔ catalog context keys stay stable across leases and change when the captured document changes (0.50225ms)
✔ binding changes after catalog preparation refuse dispatch before entering a handler (0.738875ms)
✔ context identity is rechecked after asynchronous validation before the handler starts (0.398833ms)
✔ handler assertCurrent rejects a binding change across await and a started write remains unknown (0.407541ms)
✔ handler assertCurrent also rejects its cancelled dispatch after a noncooperating await (0.30325ms)
✔ a retained execution context is revoked when a successful dispatch has returned (0.271917ms)
✔ independent dispatch cancellation bounds a hanging getContext check before_handler (0.42675ms)
✔ independent dispatch cancellation bounds a hanging guard check before_handler (0.284625ms)
✔ independent dispatch cancellation bounds a hanging getContext check after_handler (0.369541ms)
✔ independent dispatch cancellation bounds a hanging guard check after_handler (0.375333ms)
✔ stable contract hashes exclude opaque context values while including schemas and all registry tools (0.308083ms)
✔ an already established registry rejects base-contract and tenant identity changes (0.169875ms)
✔ context resolver failures expose only a fixed registry error (0.21675ms)
✔ runAgent uses catalog dispatch, preserves callId, and records unknown custom writes for reconciliation (2.442833ms)
✔ resume rejects schema, version, effect, origin, base-contract and same-origin tenant changes before planning (6.450833ms)
✔ reconciliation of an unknown write cannot rebind its checkpoint to another tenant on the same origin (1.154709ms)
✔ runAgent discards a decision when the captured browser binding changes during planning (0.754083ms)
✔ viewport observation reaches the 501st button after scrolling a truncated page (708.855625ms)
✔ scoped observation exposes the selected subtree and resets incompatible diff baselines (197.62175ms)
✔ viewport text clips individual offscreen lines within a single text node (181.063291ms)
✔ viewport observation excludes controls and text clipped by scroll-container ancestry (178.030416ms)
✔ snapshots, extraction, verification and guarded actions share open shadow and slot content (235.448792ms)
✔ hidden hosts, hidden slots and raw form values do not leak through composed readers (290.327958ms)
✔ text matching searches beyond bounded evidence and across composed text-node boundaries (177.52125ms)
✔ explicit hidden accessibility labels still name guarded controls without leaking form values (229.386208ms)
✔ link and table extraction report item and character truncation accurately (191.797208ms)
✔ scan-budget exhaustion is explicit in both snapshots and extraction (269.826833ms)
✔ ordinary draggable and aria-labeled divs deliver actual native drop payload (736.266667ms)
✔ mouse-driven custom drag supports the observed source covering its drop target (947.554875ms)
✔ nested scaled iframe drag maps main-page coordinates and preserves native payload (1468.524083ms)
✔ open Shadow DOM drag target hit testing reaches the actual source and destination (738.808042ms)
✔ an overlay appearing after the initial mouse move cannot intercept mousedown (1394.428584ms)
✔ an overlay appearing during movement cancels the drag before drop and releases the mouse (925.3375ms)
✔ mouseup RPC rejection is a failed action even if the release was applied; cleanup releases held input (1477.261334ms)
✔ an unreleasable mouse closes only the owned session and reports input cleanup failure (876.202958ms)
✔ schema extraction binds every populated leaf to exact source text and authoritative URLs (16.127916ms)
✔ multiple sources can produce nested data and a checked aggregate without a provider dependency (6.886666ms)
✔ invented sources, URLs, approximate quotes and missing or non-leaf citations are rejected (21.861958ms)
✔ JSON pointers escape literal keys and require evidence for null and empty containers (2.237541ms)
✔ schema types, required fields, formats and local references are enforced without coercion (10.614333ms)
✔ literal schema const data is not interpreted as schema references (3.326792ms)
✔ source and candidate limits fail before accepting output (14.551375ms)
✔ quote presence is distinct from truth and the application can reject unsupported claims (16.995375ms)
✔ cancellation and deadline stop waiting on non-cooperating extractors without retries (16.589041ms)
✔ deterministic fields create typed values and preserve their exact DOM provenance (1.840625ms)
✔ field-map cardinality, optional omissions, strict conversions and observation binding are enforced (21.363834ms)
✔ browser inspector uses composed text, current form state and real page provenance (641.706708ms)
✔ find traverses actual virtualized rows and returns a fresh actionable viewport ref (4852.354583ms)
✔ MCP find supports a selected iframe and preserves current snapshot semantics (838.032667ms)
✔ closing a CDP session cancels its unfinished download without closing the external browser (804.742209ms)
✔ a tab created after disposal begins is closed instead of escaping session ownership (766.553167ms)
✔ coordinate clicks never move to a sibling tab when the observed tab closes (788.405208ms)
✔ allowed native form submission preserves one POST with the actual encoded business values (753.114959ms)
✔ allowed 307 redirect preserves the original POST method and body without duplicate dispatch (739.530709ms)
✔ same-origin 302 processes Set-Cookie and a matching workspace restores that cookie and policy (1072.723334ms)
✔ explicit open/goto/new-tab deny the other port before a request or a new tab exists (748.853583ms)
✔ multi-hop HTTP 302 is intercepted before the denied hop during open, goto and new-tab (900.34875ms)
✔ iframe redirect is blocked before its denied server receives a request and counts remain per context (991.413208ms)
✔ one context blocking a document does not interrupt a different context pending business action (1216.104791ms)
✔ a new popup first-document redirect is blocked even before an ordinary popup snapshot exists (865.410417ms)
✔ page script and meta refresh cannot issue an HTTP document request to a denied origin (1087.993667ms)
✔ an allowed localhost child of the 127.0.0.1 page retains interception through its cross-site redirect (996.280041ms)
✔ non-document fetch and image requests reach the other origin: navigation policy is not a network firewall (526.739667ms)
✔ workspace policy changes, missing hashes, policy removal and newly added policy reject before browser launch (0.729459ms)
✔ CDP with configured navigation policy is rejected by the constructor without a connection attempt (52.495958ms)
✔ absent navigation policy retains ordinary browsing and does not attach a policy hash or diagnostics (590.244042ms)
✔ doctor validates and summarizes canonical policy without launching or printing origins (784.064541ms)
✔ policy file parsing rejects unsafe shapes, origin syntax, unknown fields and raw error disclosure (2897.336042ms)
✔ policy reader enforces a 64 KiB file bound and refuses directories and missing files (769.895709ms)
✔ policy reader refuses a FIFO without waiting for a writer (194.130208ms)
✔ policy and external CDP or setup are rejected before reading the policy path (770.682833ms)
✔ malformed private checkpoints do not disclose JSON excerpts before policy restoration (582.243292ms)
✔ MCP startup accepts a valid policy and exposes tools without launching a browser (208.762375ms)
✔ empty CLI checkpoints retain policy identity and reject changed, removed or legacy policy on resume (1239.30975ms)
✔ dispose closes its already-owned browser while policy bootstrap is still awaiting a CDP response (2.451959ms)
✔ an in-flight real navigation reports a safe policy fault instead of the transport URL/error (633.504583ms)
✔ policy failure after a real write retains partial action evidence and skips the next mutation (723.218708ms)
✔ undefined disables policy while configured empty policy permits only HTTP(S) (0.675375ms)
✔ explicit empty allowlist denies all targets and differs from an absent allowlist (0.316208ms)
✔ exact origin matching preserves scheme, port and host boundaries with blocked precedence (0.238ms)
✔ WHATWG normalization covers case, default ports, Unicode hostnames, IPv4 and IPv6 (0.121083ms)
✔ canonical identity is versioned, stable across equivalent ordering and does not collapse deny-all (0.210042ms)
✔ compiled policy and copied arrays cannot be mutated and detached methods retain their policy (0.382959ms)
✔ only origins are accepted as rules; URL parser repairs and wildcard guesses are rejected (1.139833ms)
✔ strict bounded configuration rejects unknown fields, wrong types, sparse arrays and accessors (0.749458ms)
✔ combined allow/block input is limited to 200 entries before normalization or deduplication (0.6105ms)
✔ invalid configuration, malformed navigation and denied URLs never surface private input (0.381041ms)
✔ opt-in MCP journal correlates an authenticated popup response without exposing request query or headers (966.631333ms)
✔ network capture is absent by default and refuses bodies without a declared bound (599.1895ms)
✔ attached CDP capture ignores independent Chrome pages (612.475958ms)
✔ response bodies remain gated after a scoped credential is filled (628.055292ms)
✔ pre-cancelled open does not start a browser or create a session (3.602791ms)
✔ cancellation during shared browser startup returns promptly and leaves it available (590.481417ms)
✔ a context that arrives after cancellation is closed without touching a sibling session (574.600166ms)
✔ page creation cancellation closes the owned context and any late page (645.753042ms)
✔ navigation cancellation removes only the opening attempt and drains its resources (704.317375ms)
✔ snapshot cancellation cannot register a completed late observation as a live session (677.001917ms)
✔ MCP tab_open propagates cancellation to the engine while sibling sessions survive (750.852042ms)
✔ CDP cancellation cleans a late owned page and leaves independent pages and browser alive (1063.107041ms)
✔ dispose is bounded while an isolated context acquisition has not returned (2430.889417ms)
✔ dispose is bounded while an isolated page acquisition has not returned (2518.416959ms)
✔ dispose reports context close failures rather than claiming complete cleanup (643.744916ms)
✔ dispose retains an existing session cleanup after close removes the session from its map (2453.694084ms)
✔ CDP disposal reports an unidentified pending page and closes it after late arrival (2726.02775ms)
✔ owned profile survives a cold browser restart, rejects concurrent use, and isolates same-origin users (2275.714834ms)
✔ an existing unrelated browser directory is never adopted as a Tablaze profile (1.611292ms)
✔ a failed Chrome launch leaves the profile reusable without exposing its path (628.508459ms)
✔ stdio MCP can reopen its owned profile with an operator-supplied identity (1757.951291ms)
✔ planner removes only exact plain duplicates while preserving errors and distinct or annotated text (9.722166ms)
✔ planner retains text-only fallback and near-duplicate serialization (0.242167ms)
✔ projection preserves image order and complete assistant/tool groups with and without image support (0.532458ms)
✔ projection leaves successful evidence, full history, events, and saved checkpoints intact (4.040625ms)
✔ default stay keeps the opener active and completes remaining actions (763.566625ms)
✔ follow-single returns popup observation and truthfully skips remaining old-ref actions (913.595459ms)
✔ multiple candidates are retained without choosing the first or stealing focus (812.54975ms)
✔ a late popup is owned but never retroactively changes the active tab (1589.965ms)
✔ a different owned background opener cannot enter the foreground action window (947.276125ms)
✔ iframe-triggered unique popup is associated with the observed owning page (908.060292ms)
✔ about:blank popups return bounded observations without waiting for eventual navigation (892.981209ms)
✔ popup collection consumes the existing batch deadline and never retries the input (643.711375ms)
✔ workspace preserves policy and old version-one records restore the stay default (2013.633083ms)
✔ failed workspace restoration rolls back policy as well as newly owned sessions (553.817667ms)
✔ stdio handshake exposes the bounded browser tools and structured session errors (211.47725ms)
✔ stdio tools drive isolated Chromium and return browser evidence (929.854417ms)
✔ SDK cancellation stops the browser batch before its later side effect (847.500875ms)
✔ CDP connection errors never return endpoint credentials (217.140166ms)
✔ a total batch budget ends a longer individual wait and reports skipped work (1000.381958ms)
✔ CLI help, version and doctor are usable without launching a browser (788.695834ms)
✔ provider selection preserves required explicit model and rejects mismatched options before execution (2879.279625ms)
✔ new provider-only options cannot start the default MCP server (767.975291ms)
✔ Anthropic CLI forwards its explicit model, native tool schema, key and output cap; retains real usage (215.599ms)
✔ Ollama CLI uses native requests, optional num_predict and no invented token counters (420.718833ms)
✔ explicit compatible provider keeps the previous endpoint and key contract (211.705084ms)
✔ Codex CLI uses the chosen executable and explicit model/effort without a model endpoint (642.740708ms)
✔ Codex executable defaults to codex on PATH without inventing a reasoning setting (668.5615ms)
✔ Codex failures expose fixed diagnostics without stderr, credentials or provider error text (633.793625ms)
✔ Codex provider integrates start-url, actual browser verification and normal cleanup (1277.824541ms)
✔ Codex checkpoint resume preserves prior budgets and still requires provider/model on each invocation (900.485167ms)
✔ bound checkpoints are rejected before any Codex process or browser restoration (198.332333ms)
✔ SIGTERM cancels an in-flight Codex planner and terminates its child (637.281541ms)
✔ anthropic: native wire request completes a verified real MCP tool loop (32.614833ms)
✔ anthropic: images and errors preserve paired repeated-name tool results without mutating history (5.125708ms)
✔ anthropic: text-only selection omits image bytes and native defaults remain explicit (1.612583ms)
✔ anthropic: malformed native decisions, mixed controls and unknown control fields fail safely (5.193959ms)
✔ anthropic: HTTP diagnostics discard private error bodies and do not retry (1.636708ms)
✔ anthropic: native response size is bounded before dropping unknown fields (1.078375ms)
✔ anthropic: cancellation interrupts a pending native response body (1.623708ms)
✔ anthropic: actual missing or invalid usage counters stay absent (1.359ms)
✔ anthropic: malformed JSON and interrupted bodies retain distinct safe response failures (54.500583ms)
✔ anthropic: unpaired history is rejected before any native request (0.502625ms)
✔ ollama: native wire request completes a verified real MCP tool loop (5.041333ms)
✔ ollama: images and errors preserve paired repeated-name tool results without mutating history (1.128625ms)
✔ ollama: text-only selection omits image bytes and native defaults remain explicit (0.733292ms)
✔ ollama: malformed native decisions, mixed controls and unknown control fields fail safely (2.74175ms)
✔ ollama: HTTP diagnostics discard private error bodies and do not retry (0.793208ms)
✔ ollama: native response size is bounded before dropping unknown fields (0.986375ms)
✔ ollama: cancellation interrupts a pending native response body (1.535792ms)
✔ ollama: actual missing or invalid usage counters stay absent (0.806833ms)
✔ ollama: malformed JSON and interrupted bodies retain distinct safe response failures (53.776666ms)
✔ ollama: unpaired history is rejected before any native request (0.222334ms)
✔ native adapters validate configuration and use only their configured/default native endpoint (0.507ms)
✔ native factory header validation never includes private names, values or keys in exceptions (0.497875ms)
✔ spawned native-provider CLI rejects an invalid environment key without echoing it to stderr (390.029875ms)
✔ real MCP form fill/save verifies current refs and visible status without another snapshot (765.959625ms)
✔ read-only ref checks preserve actionability and can read a consumed current revision (271.308041ms)
✔ ref checks support textarea, select, empty values and open shadow controls (292.148125ms)
✔ MCP and engine reject missing/ambiguous targets and a ref without snapshot_id (186.259625ms)
✔ new observations and other sessions reject old snapshot identities; unknown refs fail (356.443792ms)
✔ replaced, renamed, hidden and navigated targets are rejected without selector fallback (889.269917ms)
✔ parallel status waits cannot preserve ref evidence from a replaced node or old document (829.127958ms)
✔ a later status cannot keep a previously matching value after the same node changes (433.488958ms)
✔ iframe refs retain their frame identity and reject detached frame evidence (375.368833ms)
✔ sensitive and non-form targets fail without leaking values; evidence remains bounded (341.985042ms)
✔ queued verification waits for the action batch and shares the original readonly revision (447.596958ms)
✔ scoped credential completes a real login and never appears in normal evidence or artifacts (1031.882583ms)
✔ iframe credentials require both the frame and top-level origin to match (788.349667ms)
✔ MCP uses only the credential alias and reports login evidence without disclosing the value (2704.706208ms)
✔ doctor reports only secret metadata and never launches the configured browser (397.166333ms)
✔ malformed secret configuration fails before doctor or run with one fixed message (1162.504083ms)
✔ CDP and setup reject secrets before reading the configuration path (767.692ms)
✔ stdio accepts configured aliases and exposes the public action without resolving or launching (206.407917ms)
✔ CLI run saves secret policy identity and requires the same configuration on resume (829.7195ms)
✔ configuration does not read env values until an allowed-origin resolution and supports rotation (5.303167ms)
✔ returned configuration and scopes are frozen, with only explicit artifact opt-in (1.320083ms)
✔ resolvers reject missing, inherited, empty and oversized values without disclosing their source (1.192458ms)
✔ aborted resolution does not read env and an abort during a getter cannot return the value (0.625416ms)
✔ strict JSON shape and shared SDK metadata limits reject malformed private configurations (4.0815ms)
✔ reader enforces the actual 64 KiB boundary, UTF-8 and regular files with fixed errors (0.884417ms)
✔ reader refuses a FIFO without blocking on a writer (2.680042ms)
✔ pre-page bridge exposes only bounded metadata and fills the original control without focusing a decoy (467.154084ms)
✔ old handles fail after identical DOM replacement without filling the replacement (173.148375ms)
✔ opaque HTTP sandbox remains denied despite replacing window.origin in page script (191.719458ms)
✔ captured native primitives withstand later prototype and global replacements (154.269167ms)
✔ bridge property and operation references cannot be replaced by page code (140.244417ms)
✔ invalid bindings, values and unsupported or disabled controls reject before writing (164.713209ms)
✔ an input listener detaching the filled node reports wrote=true and does not dispatch change or fill its replacement (157.527458ms)
✔ same-URL navigation generates a fresh document identity and rejects the prior binding (157.296584ms)
✔ text projection uses lookahead before clipping and never promotes its trailing content (0.946542ms)
✔ raw boundaries remain exact when a secret resembles the replacement marker or overlaps another secret (0.858333ms)
✔ snapshot projection covers fields, options, diffs and metadata without changing reference identity (0.545292ms)
✔ link extraction preserves href-first raw allocation when redaction shrinks the last field (0.147541ms)
✔ link text and table cells redact before their field and aggregate boundaries (0.126084ms)
✔ short secrets cannot expand links or table cells past the aggregate public budget (1.763625ms)
✔ expansion followed by shrinking never reallocates the original reader budget to later fields (0.486584ms)
✔ text extraction restores the original 20000 character boundary (0.076208ms)
✔ selected metadata is copied and recursively redacted with bounded rejection (0.253166ms)
✔ metadata bounds replacement expansion as well as raw input size (44.643292ms)
✔ published and runtime MCP schemas accept only a bounded alias plus observed ref for fill_secret (28.8315ms)
✔ secret failures retain their fixed public code in structured and text MCP results (1.689375ms)
✔ SDK server rejects secret configuration with external CDP before any resolver is called (0.272333ms)
✔ undefined disables secrets; compiled facade is frozen, detached methods work, and values are private (1.870167ms)
✔ strict options reject unknown keys, accessors, prototypes and wrong optional types without invoking getters (0.572ms)
✔ strict definitions reject invalid names, versions, duplicate aliases, unknown fields and inherited resolvers (0.283833ms)
✔ arrays reject holes, extra properties and getters; alias and configured-origin limits are bounded (1.1185ms)
✔ origin lists require exact nonempty HTTP(S) origins; wildcard, credentials, path and opaque origins are rejected (0.287958ms)
✔ frame and top origins are both required, including port and scheme; scope failure never invokes resolver (0.411459ms)
✔ hash normalizes ordering, duplicates and explicit default top scope; excludes callbacks and resolved values (0.535125ms)
✔ caller mutations cannot change cloned scopes, metadata or captured resolver (0.136625ms)
✔ context is checked before resolving and again before delivering; failed delivery still redacts the resolved value (0.225458ms)
✔ resolver and context exceptions are fixed and never expose raw thrown objects or messages (0.614ms)
✔ pre-aborted calls invoke neither context nor resolver and never expose the abort reason (0.094875ms)
✔ cancellation bounds a hanging pre-context check and does not enter the resolver later (2.199833ms)
✔ cancellation bounds resolver wait, passes an aborted signal, and registers its late value while store remains alive (0.345542ms)
✔ late resolver rejection after cancellation is observed without unhandled rejection (0.221ms)
✔ cancellation bounds a hanging post-resolution context check while preserving redaction (0.216167ms)
✔ all rotated and cross-alias values remain redacted; duplicate values do not consume retention slots (1.8245ms)
✔ resolved values must be nonempty bounded well-formed strings and errors contain no returned value (0.2525ms)
✔ 128 distinct-value limit fails before delivery without dropping earlier redactions (2.097ms)
✔ 64 KiB retention limit counts UTF-8 bytes and rejects overflow atomically (9.892209ms)
✔ raw, whitespace-normalized, URI and JSON-escaped variants redact literally and contains follows the same set (0.292125ms)
✔ overlapping secrets use longest simultaneous replacement and never reprocess the replacement marker (0.155708ms)
✔ crossing and chained matches cover every sensitive character in original coordinates (0.202125ms)
✔ public aliases equal to known values are hidden without changing configured resolution or unrelated names (0.149625ms)
✔ a late cancelled value and a newer rotated value are both retained until close (1.435375ms)
✔ padding covers maximum URI and JSON forms; redaction before truncation does not reveal a secret prefix (2.973834ms)
✔ close revokes active resolution, aborts the resolver signal and prevents silent redaction bypass afterward (0.378333ms)
✔ MCP schema extraction returns typed shadow/form data and keeps observed references usable (576.011083ms)
✔ MCP preserves structured validation error codes and pointer issues without raw page values (174.7375ms)
✔ MCP rejects sensitive, hidden, raw-form and truncated fields without partial extracted data (227.983959ms)
✔ MCP extraction follows the explicitly observed frame and cites its own URL (226.375292ms)
✔ MCP bounds field counts and rejects unsupported field options before extraction (164.215708ms)
✔ real actions show a non-intercepting pointer on owned pages and rejected refs do not fake a click (3423.729375ms)
✔ operators can disable the visual pointer (627.270417ms)
✔ sessions do not add a pointer by default, even when headed (820.674416ms)
✔ popup tabs stay owned, switching invalidates references and last close ends session (599.362958ms)
✔ navigation and explicit tabs preserve session state and handle browser history (350.673792ms)
✔ upload reads an explicit local fixture and download produces the actual file (378.85725ms)
✔ native confirm and prompt use one-shot policies and disclose actual results (469.933458ms)
✔ storage state restores cookies and localStorage across independent engine lifetimes (1048.870125ms)
✔ hover and double-click execute guarded references, coordinates reject offscreen input (315.153833ms)
✔ CSS-pixel screenshots support actual canvas coordinate input (214.664459ms)
ℹ tests 447
ℹ suites 0
ℹ pass 447
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 193209.648708
