{"version":0,"repo":"eval","title":"eval — live evidence","url":"https://store-dev.agentarmies.io/eval/demo/index.html","generated_at":"2026-09-02T18:49:00Z","fresh_as_of":"2026-09-02T18:49:00Z","generation":{"store":"agentarmies-dev","manifest_sha256":"a1950b19ef87287d39b1c4d1be17ad8196470789093a22a62124c33fd4d90605","manifest_rows":9,"source":"ladder/manifest.jsonl"},"captures":[{"capture_id":"01M1HPSKFD7MSMHF6ED19EHT7P","status":"accepted","reason":"Third Austin capture, harness d3df6c6: accepted by the lead from the pixels 2026-09-02 18:4x UTC; readiness receipt real (50,000-entry buffer, 38 cell rows unchanged across two polls), twelve controls each red for its own code; the far-field slab is CLI-1, the subject's shader bug, recorded and not gating.","source_comment":"https://linear.app/agentarmies/issue/EVAL-2/eval-the-capture-harness-with-its-admissibility-verdict-and-negative#comment-1f1b5542","harness_commit":"d3df6c646c52a9279cfe23b638429d4e86fcaca5","render_api":"webgl2","renderer_string":"ANGLE (NVIDIA, Vulkan 1.4.341 (NVIDIA NVIDIA RTX PRO 4000 Blackwell SFF Edition (0x00002C33)), NVIDIA)","readiness":{"state":"known","value":{"main_started":true,"boot_loading_hidden":true,"rung_count":4,"stable_last_repaints":[{"totalMs":66.09999999403954,"chunks":256,"frames":11,"maxChunkMs":2.399999998509884,"swapMs":0.19999999552965164},{"totalMs":55.70000001043081,"chunks":256,"frames":10,"maxChunkMs":2,"swapMs":0.3999999985098839},{"totalMs":72.10000000149012,"chunks":256,"frames":13,"maxChunkMs":1.1999999955296516,"swapMs":0.5},{"totalMs":98.59999998658895,"chunks":256,"frames":16,"maxChunkMs":3.8000000044703484,"swapMs":0.19999999552965164}],"travel_overlay_present":false,"travel_overlay_text":"","last_observed_cell_count":{"ready":0,"total":9},"travel_overlay_change_sequence":23,"tile_resource_count":38,"tile_stream_stable_polls":2,"stable_poll_interval_ms":2000,"presented_frames":60}},"clean_box":{"x":350,"y":0,"width":1980,"height":1320,"derivation":"Largest conservative axis-aligned rectangle outside the frozen build's player panel, minimap/status stack, chat, ability bar, and command tray at 2560x1440. Measured HUD clearances for this build and pose are 11 px from the player panel, 5 px from chat, 2 px from the right stack, and 2 px from the ability bar/tray; remeasure every new frame.","frame_sha256":"e09ee2420735c7e1ee3eea96fb51a0a8b3b90518ae07ff8912eac1c9423df1d7"},"image_sha256":"e09ee2420735c7e1ee3eea96fb51a0a8b3b90518ae07ff8912eac1c9423df1d7","controls":[{"name":"adapter-webgl-swiftshader","expected":"ADAPTER","passed":true,"codes":["ADAPTER","STUB"]},{"name":"adapter-webgpu-lavapipe","expected":"ADAPTER","passed":true,"codes":["ADAPTER","STUB"]},{"name":"declared-api-mismatch","expected":"API","passed":true,"codes":["API","ADAPTER","STUB"]},{"name":"uniform-frame","expected":"FRAME","passed":true,"codes":["FRAME","STUB"]},{"name":"wrong-size","expected":"SIZE","passed":true,"codes":["SIZE","STUB"]},{"name":"dirty-tree","expected":"TREE","passed":true,"codes":["TREE","STUB"]},{"name":"readiness-timeout","expected":"READY","passed":true,"codes":["STUB","READY"]},{"name":"streaming-travel-overlay","expected":"READY","passed":true,"codes":["STUB","READY"]},{"name":"streaming-cell-resources","expected":"READY","passed":true,"codes":["STUB","READY"]},{"name":"visible-hud-without-clean-box","expected":"HUD","passed":true,"codes":["STUB","HUD"]},{"name":"foreign-gpu-process","expected":"SHARED_GPU","passed":true,"codes":["STUB","SHARED_GPU"]},{"name":"local-network-escape","expected":"NETWORK","passed":true,"codes":["STUB","NETWORK"]}],"legal_status":"public-ok","legal_row":"the frozen alpha","frame_url":"https://store-dev.agentarmies.io/eval/captures/01M1HPSKFD7MSMHF6ED19EHT7P/frame.png"},{"capture_id":"01M1HF76V3P4N1FK1W2JVJMXV3","status":"superseded","reason":"First Austin capture: the travel overlay was visible in-frame (EVAL-2 F1).","source_comment":"https://linear.app/agentarmies/issue/EVAL-2/eval-the-capture-harness-with-its-admissibility-verdict-and-negative#comment-d12996f3","harness_commit":"8896d2a7c7479537f45102f9b74c9389119f7f6f","render_api":"webgl2","renderer_string":"ANGLE (NVIDIA, Vulkan 1.4.341 (NVIDIA NVIDIA RTX PRO 4000 Blackwell SFF Edition (0x00002C33)), NVIDIA)","readiness":{"state":"known","value":{"main_started":true,"boot_loading_hidden":true,"rung_count":4,"stable_last_repaints":[{"totalMs":27.99999999627471,"chunks":256,"frames":5,"maxChunkMs":0.3999999985098839,"swapMs":0.4000000022351742},{"totalMs":44.19999999552965,"chunks":256,"frames":8,"maxChunkMs":1.1000000014901161,"swapMs":0.3999999985098839},{"totalMs":83.6000000052154,"chunks":256,"frames":15,"maxChunkMs":1.8999999985098839,"swapMs":0.29999999701976776},{"totalMs":32.5,"chunks":256,"frames":6,"maxChunkMs":2.100000001490116,"swapMs":0.5}],"stable_poll_interval_ms":2000,"presented_frames":61}},"clean_box":null,"image_sha256":"0a1e23f66fad30aff769de346ecf095310a5e1361c85a6205d32db5dd623add9","controls":[{"name":"adapter-webgl-swiftshader","expected":"ADAPTER","passed":true,"codes":["ADAPTER","STUB"]},{"name":"adapter-webgpu-lavapipe","expected":"ADAPTER","passed":true,"codes":["ADAPTER","FRAME","STUB"]},{"name":"declared-api-mismatch","expected":"API","passed":true,"codes":["API","ADAPTER","STUB"]},{"name":"uniform-frame","expected":"FRAME","passed":true,"codes":["API","ADAPTER","FRAME","STUB"]},{"name":"wrong-size","expected":"SIZE","passed":true,"codes":["SIZE","STUB"]},{"name":"dirty-tree","expected":"TREE","passed":true,"codes":["TREE","STUB"]},{"name":"readiness-timeout","expected":"READY","passed":true,"codes":["STUB","READY"]},{"name":"foreign-gpu-process","expected":"SHARED_GPU","passed":true,"codes":["STUB","SHARED_GPU"]},{"name":"local-network-escape","expected":"NETWORK","passed":true,"codes":["STUB","NETWORK"]}],"legal_status":"public-ok","legal_row":"the frozen alpha","frame_url":"https://store-dev.agentarmies.io/eval/captures/01M1HF76V3P4N1FK1W2JVJMXV3/frame.png"},{"capture_id":"01M1HGBD4PXHNCAVW54R30TP9M","status":"withdrawn","legal_status":"public-ok","legal_row":"the frozen alpha","reason":"Round-one review O1: readiness trusted a saturated 250-entry resource buffer, so tile quiescence could not fail.","source_comment":"https://linear.app/agentarmies/issue/EVAL-2/eval-the-capture-harness-with-its-admissibility-verdict-and-negative#comment-5e28d76a","harness_commit":"f6b1ee11506604e3c948a90728aba8a51e2fff1d","render_api":"webgl2","renderer_string":"ANGLE (NVIDIA, Vulkan 1.4.341 (NVIDIA NVIDIA RTX PRO 4000 Blackwell SFF Edition (0x00002C33)), NVIDIA)","readiness":{"state":"known","value":{"main_started":true,"boot_loading_hidden":true,"rung_count":4,"stable_last_repaints":[{"totalMs":70.49999999627471,"chunks":256,"frames":12,"maxChunkMs":1.800000000745058,"swapMs":0.19999999925494194},{"totalMs":57.69999999552965,"chunks":256,"frames":10,"maxChunkMs":1.300000000745058,"swapMs":0.30000000074505806},{"totalMs":78.09999999776483,"chunks":256,"frames":13,"maxChunkMs":2.7000000029802322,"swapMs":0.5},{"totalMs":101.5,"chunks":256,"frames":16,"maxChunkMs":3.599999997764826,"swapMs":0.3999999985098839}],"travel_overlay_present":false,"travel_overlay_text":"","last_observed_cell_count":{"ready":1,"total":9},"travel_overlay_change_sequence":14,"tile_resource_count":4,"tile_stream_stable_polls":2,"stable_poll_interval_ms":2000,"presented_frames":60}},"clean_box":{"x":350,"y":0,"width":1980,"height":1320,"derivation":"Largest conservative axis-aligned rectangle outside the frozen build's player panel, minimap/status stack, chat, ability bar, and command tray at 2560x1440.","frame_sha256":"639096900f531cb70fa164595557919ebce8d127b2c9bd32a5c269ad7ff8d8d6"},"image_sha256":"639096900f531cb70fa164595557919ebce8d127b2c9bd32a5c269ad7ff8d8d6","controls":[{"name":"adapter-webgl-swiftshader","expected":"ADAPTER","passed":true,"codes":["ADAPTER","STUB"]},{"name":"adapter-webgpu-lavapipe","expected":"ADAPTER","passed":true,"codes":["ADAPTER","STUB"]},{"name":"declared-api-mismatch","expected":"API","passed":true,"codes":["API","ADAPTER","STUB"]},{"name":"uniform-frame","expected":"FRAME","passed":true,"codes":["FRAME","STUB"]},{"name":"wrong-size","expected":"SIZE","passed":true,"codes":["SIZE","STUB"]},{"name":"dirty-tree","expected":"TREE","passed":true,"codes":["TREE","STUB"]},{"name":"readiness-timeout","expected":"READY","passed":true,"codes":["STUB","READY"]},{"name":"streaming-travel-overlay","expected":"READY","passed":true,"codes":["STUB","READY"]},{"name":"foreign-gpu-process","expected":"SHARED_GPU","passed":true,"codes":["STUB","SHARED_GPU"]},{"name":"local-network-escape","expected":"NETWORK","passed":true,"codes":["STUB","NETWORK"]}],"frame_url":"https://store-dev.agentarmies.io/eval/captures/01M1HGBD4PXHNCAVW54R30TP9M/frame.png"},{"capture_id":"01M1HHTZXK87YM6EMVCDMEC6H8","status":"subject-evidence","legal_status":"public-ok","legal_row":"Khronos glTF Sample Assets used as eval's rendering fixtures","reason":"EVAL-5 WebGPU asset-viewer subject; frame and sidecar are published, with no proof object.","source_comment":"https://linear.app/agentarmies/issue/EVAL-5","harness_commit":"83d98f5d0efd4e82924aa20124d5ecd9cea1a5ec","render_api":"webgpu","renderer_string":"nvidia | blackwell","readiness":{"state":"known","value":{"status":"ready","stage":"ready","error":null,"asset_digest":"d4f567186fea262819aac6664cb9d7cef7f4600406038046771a69e10363935b","viewer_commit":"83d98f5d0efd4e82924aa20124d5ecd9cea1a5ec","textures_total":11,"textures_ready":11,"first_frame_presented":true,"presented_frames_after_load":9,"required_presented_frames":8,"decoder":{"ktx2":"npm-esm-main-thread","meshopt":"npm-local-script"}}},"clean_box":null,"image_sha256":"aea4cb6e317419d0fe26722eabcb47438c5ba7ef0acadc8059eed9817eb8f693","controls":[],"frame_url":"https://store-dev.agentarmies.io/eval/captures/01M1HHTZXK87YM6EMVCDMEC6H8/frame.png"}],"ladder":[{"id":"gta5-03368705c52b","rung":"gta5-ground","width":1920,"height":1080,"acquired":"2026-09-02","origin":"GearNuke capture with site watermark, 'New Gen' caption, minimap and health bar","source":{"kind":"corpus"},"bar_for":"Eye-height third-person street shot in a low-rise commercial block: sidewalk grime, chain-link, a brick wall and a paved yard as the bar for near-field material density; the protagonist and HUD are outside clean_box.","pairable":false,"clean_box":{"x":1060,"y":0,"w":860,"h":960,"sha256":"03368705c52bce6e217d7b8d700951f43c06e9df1a3b9ab7ed4b9c282bb750f3"},"legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"},{"id":"gta5-a677915bef6c","rung":"gta5-ground","width":1280,"height":720,"acquired":"2026-09-02","origin":"YouTube thumbnail composite with PlayStation/PS5 banner and 4K HDR badge","source":{"kind":"corpus"},"bar_for":"Eye-height third-person boulevard shot: lawn, parking meters, a parked sedan and towers behind palms as the bar for near-field richness and midday lighting; the protagonist and banners are outside clean_box.","pairable":false,"clean_box":{"x":675,"y":0,"w":605,"h":720,"sha256":"a677915bef6c51e6a2fa64d49be63bb32c8abdf65b964f5a4bb0d897c4fff291"},"legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"},{"id":"gta5-9506aac2c455","rung":"gta5-ground","width":1500,"height":843,"acquired":"2026-09-02","origin":"in-game capture with minimap and a '0 deaths' overlay","source":{"kind":"corpus"},"bar_for":"Elevated eye-height shot over a desert valley: far-field terrain, sparse trees and a road grid as the bar for distant placement density and haze; the protagonist and HUD are outside clean_box.","pairable":false,"clean_box":{"x":0,"y":0,"w":1500,"h":370,"sha256":"9506aac2c455e025c14e8daa42eb7cd8dfce0ea806bd1d09bcb82343d4a385ba"},"legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"},{"id":"gta5-2318b3499d14","rung":"gta5-ground","width":1080,"height":607,"acquired":"2026-09-02","origin":"clean in-game capture, no HUD or overlay","source":{"kind":"corpus"},"bar_for":"Eye-height forest trail with no character in frame: needle litter, flowering understorey and backlit conifers as the bar for ground-level material density and golden-hour lighting.","pairable":true,"clean_box":{"x":0,"y":0,"w":1080,"h":607,"sha256":"2318b3499d14270b53d5f115b55524028169c4454fe74d73e7260d3b12f25cf8"},"legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"},{"id":"gta5-1dea1704b233","rung":"gta5-ground","width":1100,"height":619,"acquired":"2026-09-02","origin":"press-style capture with a 'grand theft auto V' logo paste bottom-right","source":{"kind":"corpus"},"bar_for":"Elevated beach shot with no protagonist: stilt houses, rocks, crowded sand and surf as the bar for placement density, water contact and midday lighting; the logo paste is outside clean_box.","pairable":true,"clean_box":{"x":0,"y":0,"w":960,"h":619,"sha256":"1dea1704b233422344cd8d51aa126bd9de23c6298cb7adbee13e4a4812fe44c6"},"legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"},{"id":"gta5-57421177e559","rung":"gta5-ground","width":1280,"height":720,"acquired":"2026-09-02","origin":"in-game capture with HUD icons and a dark panel bottom-left","source":{"kind":"corpus"},"bar_for":"Eye-height third-person intersection shot: wet asphalt, a crosswalk, a parked coupe and a plaza facade as the bar for near-field road material and low-sun lighting; the protagonist and HUD are outside clean_box.","pairable":false,"clean_box":{"x":775,"y":0,"w":505,"h":720,"sha256":"57421177e55944810274934d062904d2722fb721961bffb109119ed08ff405c0"},"legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"},{"id":"capture-01m1hf76v3p4n1fk1w2jvjmxv3","rung":"alpha-frozen","sha256":"0a1e23f66fad30aff769de346ecf095310a5e1361c85a6205d32db5dd623add9","bytes":3548900,"width":2560,"height":1440,"mime":"image/png","source":{"kind":"capture","capture_id":"01M1HF76V3P4N1FK1W2JVJMXV3","build_id":"index-CO0dL_A2.js+health:dev","pose_id":"austin/ground","key":"eval/captures/01M1HF76V3P4N1FK1W2JVJMXV3/frame.png"},"acquired":"2026-09-02","licence":"own-work","servable":true,"bar_for":"The floor: the frozen alpha (build index-CO0dL_A2.js+health:dev) captured by the harness at austin/ground; every later build at this pose is judged against it.","place_id":"austin","pose_id":"austin/ground","legal_status":"public-ok","legal_row":"the frozen alpha","store":"dev","object_url":"https://store-dev.agentarmies.io/eval/ladder/sha256/0a1e23f66fad30aff769de346ecf095310a5e1361c85a6205d32db5dd623add9"},{"id":"retract-1-capture-01m1hf76v3p4n1fk1w2jvjmxv3","retracts":"capture-01m1hf76v3p4n1fk1w2jvjmxv3","reason":"Frame shows the client's travel overlay (3/9 cells ready) and the HUD; captured before EVAL-2's readiness receipt watched the cell count. Mechanism proven; row withdrawn. Lead, EVAL-2 comment 16:35Z."},{"id":"capture-01m1hpskfd7msmhf6ed19eht7p","rung":"alpha-frozen","sha256":"e09ee2420735c7e1ee3eea96fb51a0a8b3b90518ae07ff8912eac1c9423df1d7","bytes":3591175,"width":2560,"height":1440,"mime":"image/png","source":{"kind":"capture","capture_id":"01M1HPSKFD7MSMHF6ED19EHT7P","build_id":"index-CO0dL_A2.js+health:dev","pose_id":"austin/ground","key":"eval/captures/01M1HPSKFD7MSMHF6ED19EHT7P/frame.png"},"hud":"visible","clean_box":{"x":350,"y":0,"w":1980,"h":1320,"sha256":"e09ee2420735c7e1ee3eea96fb51a0a8b3b90518ae07ff8912eac1c9423df1d7"},"acquired":"2026-09-02","captured":"2026-09-02","licence":"own-work","servable":true,"legal_status":"public-ok","legal_row":"the frozen alpha","store":"dev","bar_for":"The floor: the frozen alpha (build index-CO0dL_A2.js+health:dev) captured by the harness at austin/ground; every later build at this pose is judged against it.","place_id":"austin","pose_id":"austin/ground","notes":"Frozen with 45 shader failures, ten aborted material and vehicle fetches, and readiness 0 of 9 cells at the last observation.","object_url":"https://store-dev.agentarmies.io/eval/ladder/sha256/e09ee2420735c7e1ee3eea96fb51a0a8b3b90518ae07ff8912eac1c9423df1d7"}],"judge_bundles":[{"key":"2de36934da1a","source_capture_id":"01M1HF76V3P4N1FK1W2JVJMXV3","subjects":[{"id":"capture-01m1hf76v3p4n1fk1w2jvjmxv3","legal_status":"public-ok","legal_row":"the frozen alpha"}],"status":"superseded","reason":"Superseded wave: built from the first Austin capture 01M1HF76V3P4N1FK1W2JVJMXV3 (itself superseded); EVAL-4 rebuilds every session from the accepted third capture 01M1HPSKFD7MSMHF6ED19EHT7P.","visibility":"public","bundle_prefix":"eval/judge/2de36934da1a/","instructions_url":"https://store-dev.agentarmies.io/eval/judge/2de36934da1a/bundle/instructions.md","key_prefix":"eval/judge-keys/2de36934da1a/","key_url":"https://store-dev.agentarmies.io/eval/judge-keys/2de36934da1a/key.json"},{"key":"4468ba057616","source_capture_id":"01M1HF76V3P4N1FK1W2JVJMXV3","subjects":[{"id":"capture-01m1hf76v3p4n1fk1w2jvjmxv3","legal_status":"public-ok","legal_row":"the frozen alpha"},{"id":"gta5-2318b3499d14","legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"}],"status":"superseded","reason":"Superseded wave: built from the first Austin capture 01M1HF76V3P4N1FK1W2JVJMXV3 (itself superseded); EVAL-4 rebuilds every session from the accepted third capture 01M1HPSKFD7MSMHF6ED19EHT7P.","visibility":"internal"},{"key":"a7bd1cd4353e","source_capture_id":"01M1HF76V3P4N1FK1W2JVJMXV3","subjects":[{"id":"capture-01m1hf76v3p4n1fk1w2jvjmxv3","legal_status":"public-ok","legal_row":"the frozen alpha"}],"status":"superseded","reason":"Superseded wave: built from the first Austin capture 01M1HF76V3P4N1FK1W2JVJMXV3 (itself superseded); EVAL-4 rebuilds every session from the accepted third capture 01M1HPSKFD7MSMHF6ED19EHT7P.","visibility":"public","bundle_prefix":"eval/judge/a7bd1cd4353e/","instructions_url":"https://store-dev.agentarmies.io/eval/judge/a7bd1cd4353e/bundle/instructions.md","key_prefix":"eval/judge-keys/a7bd1cd4353e/","key_url":"https://store-dev.agentarmies.io/eval/judge-keys/a7bd1cd4353e/key.json"},{"key":"c6ff63275862","source_capture_id":"01M1HF76V3P4N1FK1W2JVJMXV3","subjects":[{"id":"capture-01m1hf76v3p4n1fk1w2jvjmxv3","legal_status":"public-ok","legal_row":"the frozen alpha"},{"id":"gta5-1dea1704b233","legal_status":"internal","legal_row":"GTA5 reference captures (the ladder's rungs)"}],"status":"superseded","reason":"Superseded wave: built from the first Austin capture 01M1HF76V3P4N1FK1W2JVJMXV3 (itself superseded); EVAL-4 rebuilds every session from the accepted third capture 01M1HPSKFD7MSMHF6ED19EHT7P.","visibility":"internal"}],"austin_verdict":"Current Austin verdict: accepted capture 01M1HPSKFD7MSMHF6ED19EHT7P.","known_conditions":["01M1HPSKFD7MSMHF6ED19EHT7P is the accepted Austin capture (Third Austin capture, harness d3df6c6: accepted by the lead from the pixels 2026-09-02 18:4x UTC; readiness receipt real (50,000-entry buffer, 38 cell rows unchanged across two polls), twelve controls each red for its own code; the far-field slab is CLI-1, the subject's shader bug, recorded and not gating.); the earlier Austin frames below stay visible as superseded or withdrawn.","Frozen alpha CLI-1: a fragment shader fails to compile and flattens the far field.","The superseded EVAL-4 judge wave must be rebuilt; its published key receipts name 01M1HF76V3P4N1FK1W2JVJMXV3, while the wave brief names 01M1HGBD4PXHNCAVW54R30TP9M.","GTA5 is a look rung for material density, near-field richness, and lighting; it is not pose-matched Austin geometry."]}