Files
Agent-Workbench/docs/turn-progress-b-evidence/logs/shadow-related.tap
T

77 lines
8.2 KiB
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
✔ reviewed application invalidates old checks, names current SHA and advances through a new approved check to Git preview (14934.2695ms)
✔ skip-tests, foreign task, unknown operations, diagnostic and parent review retain their boundaries (5998.3429ms)
✔ before the first change, real baseline failures persist across later reads; passed baselines and changed sources cannot certify delivery (8726.5505ms)
✔ unchanged proposals cannot hide repeated original-log reads or identical session inspections (13890.2694ms)
✔ reusing a verified proof never resets identical-state detection or creates a new proof (11430.2849ms)
✔ coverage merges adjacent pages, separates changed sources and never claims old Run coverage in a continuation (0.8967ms)
✔ continuations distinguish unread source, later original line ranges and already returned result wrappers (919.8288ms)
✔ pure-file batch returns unused small-page space while mixed and historical allocation stays conservative (0.3083ms)
✔ observed upstream high input pauses before a new request, preserves reported usage and distinguishes the limiting ledger (1033.2274ms)
✔ chat-completions: complete actual Unicode/CRLF pages and small source survive the next bounded wire request (3208.6141ms)
✔ responses: complete actual Unicode/CRLF pages and small source survive the next bounded wire request (2108.9623ms)
✔ anthropic-messages: complete actual Unicode/CRLF pages and small source survive the next bounded wire request (2380.3449ms)
✔ requirements HTTP binds selection, enforces optimistic revisions, restores original proofs and leaves provider history untouched (5124.8945ms)
✔ research evidence binds actual original bytes; model completion is not user acceptance (1525.6296ms)
✔ same text with distinct user source instances stays distinct; an appended requirement only invalidates its own revision (1845.1698ms)
✔ foreign project, old Task and fabricated evidence strings are rejected with preserved failed proofs (2178.0677ms)
✔ source changes invalidate prior proofs and cannot certify a read that was already stale at binding (1707.8169ms)
✔ missing or tampered artifacts never become evidence and original failures remain after recovery (1191.2175ms)
✔ checks omitted by explicit agreement are reported as not executed and never passed (811.3584ms)
✔ user confirmation stays separate from engineering evidence, is bounded to current inputs and cannot be silently converted (994.31ms)
✔ existing multi-file results keep provenance and invalidate only evidence whose files changed (1310.6476ms)
✔ unrelated additions preserve the checked file facts but remain visible as unreviewed user sources (1414.4197ms)
✔ actual Git preview and publication recheck changed requirement bindings without touching HEAD or the owner index (6774.0966ms)
✔ no-test file application and manual completion keep pending user requirements for explicit review (1547.9085ms)
✔ revision races, forged source ranges, evidence fields and deletion cannot overwrite sealed requirements (878.6389ms)
✔ requirement binding prevents completion and Git acceptance after a review changes; legacy tasks stay explicit (2695.488ms)
✔ a stop from the event loop interrupts immediate model replies before later tools (1460.0563ms)
✔ a stop after the first tool preserves a complete interrupted multi-call protocol group (1828.7444ms)
✔ the compatible HTTP loop preserves call IDs, validates errors and waits before executing commands (5278.646ms)
✔ denial prevents command execution and produces a paired denied result (1923.9518ms)
✔ stopping during approval interrupts pending calls without executing or replaying them (1951.3856ms)
✔ stopping a running fixture command terminates its own process tree and retains uncertain state (2086.0807ms)
✔ command output decodes split UTF-8 and excludes the workbench key from child environment (1010.3875ms)
✔ model-list testing uses a tentative key without changing the saved connection and redacts service errors (6.5565ms)
✔ truncated or repeated model calls fail before execution; tool input validation rejects unknown capabilities (2133.6342ms)
✔ request projection keeps exact tool pairs and complete recent user instructions while preserving disk history (1064.535ms)
✔ fixed normal exploration, pagination, SHA changes and repeated ranges are external observations, not a score (1.3157ms)
✔ read-only research is valid work; model summaries and unchanged timestamped listings do not become engineering evidence (0.3887ms)
✔ new permissions, network and broken tools are failure evidence; repeated invalid inputs remain candidates (0.3723ms)
✔ waiting for original long checks, review and approvals stays waiting across repeated observations (0.4818ms)
✔ baseline facts, real repairs, current-source check results and requirement evidence have distinct growth (0.2454ms)
✔ bounded memory keeps unknown instead of claiming an unretained range is repeated; Task/Run instances cannot mix (12.1744ms)
✔ actual runtime records growing pages then a repeated page without changing model rounds, stop rules or original tool pairs (5054.6617ms)
✔ actual invalid tool calls repeat under the original step budget; observation never adds an early stop (4684.7063ms)
✔ actual command approval is observed as waiting and denial stays a permission failure (3097.2297ms)
✔ network errors remain model failures with shadow failure evidence, never a fabricated no-progress cause (4430.238ms)
✔ observer receipt failure cannot change tool results, model budget or completion (2315.3024ms)
✔ a later receipt recovers after an observation write failure without misaligning retained source receipts (3580.1383ms)
✔ real applied files, counted checks and current requirement proofs grow after an existing-facts baseline (4019.495ms)
✔ receipt retention stays aligned at the cap and a foreign step cannot enter the Run (3120.7675ms)
✔ all actually returned same-version fragments reach the bounded request, including overlap and Unicode, with independent original provenance (5030.7216ms)
✔ unread gaps and different range origins remain separate; partial sending names the precise missing already-observed range (923.3792ms)
✔ same-SHA conflicting overlaps and invalid coordinates fail closed, while a new SHA never inherits old fragments (1424.9027ms)
✔ versioned old views keep the original single-page semantics and verification (1119.7532ms)
✔ small code and actual failed command stay visible when their original multi-tool group cannot fit (1274.8968ms)
✔ mixed group retains six distinct small file observations with exact multilingual text and bounded request bytes (1891.5719ms)
✔ working observations are bound to the current Task and stop treating a pre-application file revision as current evidence (2336.5661ms)
✔ a sealed observation rejects changed or missing original result artifacts instead of recreating evidence (1473.7606ms)
✔ a real passed-check observation is excluded after a source edit invalidates the check, while its historical sent evidence remains valid (4656.689ms)
✔ budgeted file pages reconstruct UTF-8 text including long single lines and surrogate boundaries (4.6712ms)
✔ actual HTTP multi-tool reads survive an oversized runtime snapshot and the sealed evidence survives reopening (9768.7867ms)
✔ actual HTTP file pagination advances over a long file with each returned page visible in the next request (6083.3053ms)
✔ later mixed history pages cannot displace the actual module sources and complete requirements (4348.3735ms)
✔ a later narrow page of the same file revision cannot evict the wider actual observation (2772.2176ms)
✔ six small source files stay complete across metadata turns even when mandatory evidence exceeds the soft target (2776.0626ms)
✔ historical read-artifact navigation points to the original artifact and next page, never its result wrapper (978.529ms)
✔ a 1.6.1 full-provenance sent view remains verifiable after introducing the compact view (977.3426ms)
✔ actual HTTP requests keep four observed modules after a later large history read and preserve all source bytes (9659.4196ms)
ℹ tests 68
ℹ suites 0
ℹ pass 68
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 117743.4888