coverage: what the solvers could not answer at release, day by day

autopilot claims every single at the 09:00 et release and runs the lane's newest solver on the real bytes. when the solver could not answer a field, autopilot posted a gaps row naming the field families it missed and the crash behind them, and an agent took the single from there. this page reads those rows lane by lane and day by day, beside what each lane's tooling gained that night. days 1 to 4 ran before the release pass existed.

  • 155non-web singles raced5 lanes, 31 days
  • 89covered at release66% of the 135 from day 5 on
  • 39partiala gaps row, flagged inside 3 min
  • 5cold startsa gaps row, 3 to 15 min
  • 1long daysa gaps row, over 15 min
  • 20before the release passdays 1 to 4; 1 more unknown
  • warmupsbest coverage share24 of 27 covered, days 5 to 31
  • scriptmost improved lanecovered 69% days 5 to 17, 100% days 18 to 31

the month at a glance

a cell's colour follows one rule. covered: autopilot submitted the single and posted no gaps row for it. partial: autopilot posted a gaps row and the single was still flagged within 180 s of the 09:00 et release. cold start: a gaps row and 180 s to 15 min. long: a gaps row and over 15 min.

each lane's single at release, by day
12345678910111213141516171819202122232425262728293031warmupsday 1 warmups: no release pass, single at 2h30m (unattributed). no release pass yet: autopilot's first release pass was day 5day 2 warmups: no release pass, single at 10:25 (helper). no release pass yet: autopilot's first release pass was day 5day 3 warmups: no release pass, single at 0:26 (helper). no release pass yet: autopilot's first release pass was day 5day 4 warmups: no release pass, single at 0:58 (helper). no release pass yet: autopilot's first release pass was day 5 (on day 4 it first ran at 11:44:40, after the singles)day 5 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 6 warmups: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 7 warmups: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 8 warmups: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 9 warmups: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 10 warmups: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 11 warmups: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 12 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 13 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 14 warmups: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 15 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 16 warmups: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 17 warmups: partial, single at 0:42 (lane fixer). 1 field (message): the dry run ran past its 15 s capday 18 warmups: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 19 warmups: partial, single at 0:26 (parent). 1 field (message): no decoding path foundday 20 warmups: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 21 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 22 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 23 warmups: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 24 warmups: partial, single at 0:21 (lane fixer). 1 field (message): no decoding path foundday 25 warmups: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 26 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 27 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 28 warmups: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 29 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 30 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 31 warmups: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releasemetadataday 1 metadata: no release pass, single at 2h30m (unattributed). no release pass yet: autopilot's first release pass was day 5day 2 metadata: no release pass, single at 9:06 (parent). no release pass yet: autopilot's first release pass was day 5day 3 metadata: no release pass, single at 0:28 (parent). no release pass yet: autopilot's first release pass was day 5day 4 metadata: no release pass, single at 1:12 (helper). no release pass yet: autopilot's first release pass was day 5 (on day 4 it first ran at 11:44:40, after the singles)day 5 metadata: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 6 metadata: partial, single at 1:04 (helper). 1 field (d06-pdf): no reader for d06-pdf-001-pdf-header-versionday 7 metadata: partial, single at 1:05 (helper). 1 field (d07-zip): no reader for zip-eocd-total-entriesday 8 metadata: partial, single at 1:06 (helper). 2 fields (d07-zip, evidence-note): no reader for zip-central-directory-member-namesday 9 metadata: partial, single at 0:26 (helper). 1 field (d09-pe): no reader for pe-coff-machine-u16-hexday 10 metadata: partial, single at 1:08 (helper). 1 field (d09-pe): no reader for pe-address-of-entry-point-rva-hexday 11 metadata: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 12 metadata: partial, single at 1:03 (helper). 1 field (d12-dotnet): no reader for dotnet-mvid-guidday 13 metadata: partial, single at 1:26 (parent). 1 field (d13-msi): no reader for msi-product-code-guidday 14 metadata: partial, single at 0:18 (helper). 1 field (d14-lnk): no reader for lnk-target-path-windowsday 15 metadata: partial, single at 0:23 (lane fixer). 1 field (d15-ooxml): no reader for ooxml-external-relationship-targetsday 16 metadata: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 17 metadata: partial, single at 0:18 (lane fixer). 1 field (d17-audio): no reader for media-primary-audio-codec-nameday 18 metadata: partial, single at 0:18 (lane fixer). 1 field (d18-video): KeyError: "media-primary-video-dimensions: the video stream has no 'dimensions'"day 19 metadata: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 20 metadata: partial, single at 0:38 (lane fixer). 1 field (d20-macho): no reader for macho-cpu-type-setday 21 metadata: partial, single at 0:30 (lane fixer). 1 field (d21-apk): no reader for apk-package-nameday 22 metadata: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 23 metadata: partial, single at 0:15 (autopilot). 1 field (d23-sqlite): no crash; the solver's 12288 was rejectedday 24 metadata: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 25 metadata: partial, single at 0:18 (autopilot). 1 field (d25-registry): no reader for registry-hive-root-nameday 26 metadata: partial, single at 1:28 (autopilot). 1 field (d26-prefetch): no reader for prefetch-executable-nameday 27 metadata: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 28 metadata: cold start, single at 8:16 (lane fixer). 1 field (d28-disk): no reader for disk-partition-start-lba-sequenceday 29 metadata: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 30 metadata: partial, single at 0:56 (autopilot). 1 field (d30-cluster): no reader for relation-identical-sha256-groupsday 31 metadata: partial, single at 1:11 (lane fixer). 1 field (d31-case): no reader for graph-execution-chain-artifact-idsmobileday 1 mobile: no release pass, single at 2h48m (unattributed). no release pass yet: autopilot's first release pass was day 5day 2 mobile: no release pass, single at 9:56 (lane fixer). no release pass yet: autopilot's first release pass was day 5day 3 mobile: no release pass, single at 0:30 (helper). no release pass yet: autopilot's first release pass was day 5day 4 mobile: no release pass, single at 0:49 (helper). no release pass yet: autopilot's first release pass was day 5 (on day 4 it first ran at 11:44:40, after the singles)day 5 mobile: partial, single at 1:16 (helper). 1 field (answer): no decoding path foundday 6 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 7 mobile: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 8 mobile: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 9 mobile: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 10 mobile: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 11 mobile: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 12 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 13 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 14 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 15 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 16 mobile: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 17 mobile: partial, single at 0:19 (lane fixer). 1 field (answer): no decoding path foundday 18 mobile: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 19 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 20 mobile: partial, single at 0:02 (autopilot). 1 field (answer): no decoding path foundday 21 mobile: covered, single at 0:01 (autopilot). nothing: autopilot answered every field at releaseday 22 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 23 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 24 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 25 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 26 mobile: partial, single at 0:29 (parent). 1 field (answer): the split parts did not joinday 27 mobile: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 28 mobile: cold start, single at 14:12 (parent). 1 field (answer): no decoding path foundday 29 mobile: partial, single at 0:17 (lane fixer). 1 field (answer): no decoding path foundday 30 mobile: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 31 mobile: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releasescriptday 1 script: no release pass, single at 2h45m (unattributed). no release pass yet: autopilot's first release pass was day 5day 2 script: no release pass, single at 12:35 (helper). no release pass yet: autopilot's first release pass was day 5day 3 script: no release pass, single at 7:18 (parent). no release pass yet: autopilot's first release pass was day 5day 4 script: no release pass, single at 3:27 (helper). no release pass yet: autopilot's first release pass was day 5 (on day 4 it first ran at 11:44:40, after the singles)day 5 script: cold start, single at 3:29 (helper). 126 fields in 10 families (python): ValueError: no Material_ const in vbscript; ValueError: vbscript material does not decodeday 6 script: partial, single at 1:27 (parent). 369 fields in 58 families (bash, batch, collection, impact, jscript, python, …): no crash; 5 solver answers rejected, e.g. decode_configuration,return_configurati…day 7 script: partial, single at 2:25 (parent). 255 fields in 32 families (bash, nodejs, python, resource, vbscript): a KeyError on persistence-adapter, persistence-attack, persistence-campaignday 8 script: partial, single at 2:07 (helper). 313 fields in 28 families (host, polyglot, python, resource, vbscript): no answer for host-php-webshell-php-obfuscation, polyglot-php-benign-utility-obfuscation, polyglot-php-launcher-obfuscationday 9 script: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 10 script: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 11 script: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 12 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 13 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 14 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 15 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 16 script: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 17 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 18 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 19 script: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 20 script: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 21 script: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 22 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 23 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 24 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 25 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 26 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 27 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 28 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 29 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 30 script: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 31 script: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releasebinaryday 1 binary: no release pass, single at 2h43m (unattributed). no release pass yet: autopilot's first release pass was day 5day 2 binary: no release pass, single at 7:10 (parent). no release pass yet: autopilot's first release pass was day 5day 3 binary: no release pass, single at 9:37 (parent). no release pass yet: autopilot's first release pass was day 5day 4 binary: no release pass, single at 2:31 (helper). no release pass yet: autopilot's first release pass was day 5 (on day 4 it first ran at 11:44:40, after the singles)day 5 binary: unknown, single at 0:28 (helper). no gaps row: autopilot's pass on it failed before its dry run (ModuleNotFoundError: No module named 'pefile')day 6 binary: partial, single at 1:40 (helper). 8 fields (d06-analysis, d06-artifact): a KeyError on 06day 7 binary: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 8 binary: partial, single at 1:54 (helper). 11 fields in 4 families (d02, d06, d07, d08): no producer for compression_methods, has_cpp_typeinfo, has_cpp_vtableday 9 binary: partial, single at 1:03 (helper). 1 field (d09-analysis): no producer for hcfg_offsetday 10 binary: partial, single at 1:02 (helper). 1 field (d10-analysis): no producer for has_hcfg_sectionday 11 binary: covered, single at 0:02 (autopilot). nothing: autopilot answered every field at releaseday 12 binary: covered, single at 0:06 (autopilot). nothing: autopilot answered every field at releaseday 13 binary: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 14 binary: covered, single at 0:05 (autopilot). nothing: autopilot answered every field at releaseday 15 binary: partial, single at 1:02 (lane fixer). 8 fields (d15-analysis): no producer for stripped_architecture, stripped_config_campaign, stripped_config_domainday 16 binary: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 17 binary: long, single at 1h41m (helper). 12 fields in 11 families (d02, d04, d05, d06, d07, d08, …): no producer for access_flags, has_coroutine_runtime, has_go_runtimeday 18 binary: covered, single at 0:03 (autopilot). nothing: autopilot answered every field at releaseday 19 binary: partial, single at 0:54 (autopilot). 8 fields (d19-analysis): no producer for overlay_offset, overlay_size; no valid hcfg blob in any memberday 20 binary: partial, single at 0:42 (autopilot). 1 field (d20-analysis): no producer for hcfg_resource_sizeday 21 binary: partial, single at 0:22 (autopilot). 6 fields (d21-analysis): no producer for loader_machine, plugin_config_campaign, plugin_config_domainday 22 binary: covered, single at 0:08 (autopilot). nothing: autopilot answered every field at releaseday 23 binary: covered, single at 0:04 (autopilot). nothing: autopilot answered every field at releaseday 24 binary: covered, single at 0:07 (autopilot). nothing: autopilot answered every field at releaseday 25 binary: covered, single at 0:06 (autopilot). nothing: autopilot answered every field at releaseday 26 binary: partial, single at 1:02 (autopilot). 2 fields (d26-analysis): no producer for crash_fault_method, crash_frames_match_methodsday 27 binary: covered, single at 0:09 (autopilot). nothing: autopilot answered every field at releaseday 28 binary: cold start, single at 13:37 (parent). 32 fields in 16 families (d01, d03, d05, d06, d09, d10, …): no producer for class_name, cli_runtime, conditional_branch_opcode_countday 29 binary: covered, single at 0:07 (autopilot). nothing: autopilot answered every field at releaseday 30 binary: cold start, single at 3:58 (autopilot). 8 fields (d30-analysis): no valid hcfg blob in any member; an ungraded binary format, neko_bytecodeday 31 binary: partial, single at 1:08 (lane fixer). 10 fields (d31-analysis): no producer for config_campaign, config_domain, config_mutexwebday 19 web: no release pass, single at 2:32 (web worker). no web autopilot pass ran this day: the release pass's web listing returned no rows and no autopilot intel row exists for the day; all three singles were flagged by a web worker from the homelab vmday 20 web: cold start, single at 6:02 (parent). the matcher left proof-1 unanswered on lfi at release (no request or static source rule for proof-1); administratively-speaking went unattributed without an autopilot pass; lfi was flagged by the parentday 21 web: long, single at 23:04 (unattributed). the matcher left proof-1 unanswered on session-impression, stored-xss and reflect-upon-this at release (the capture held no proof); all three singles went unattributedday 22 web: cold start, single at 8:08 (unattributed). the matcher left proof-1 unanswered on forged-with-care and uploading-expectations at release (the capture held no proof); uploading-expectations went unattributed; forged-with-care was flagged by a web workerday 23 web: cold start, single at 4:40 (unattributed). the matcher left proof-1 unanswered on ask-again-later, template-temper and posted-too-much at release (the capture held no proof); all three singles went unattributedday 24 web: partial, single at 1:54 (unattributed). the matcher left proof-1 unanswered on fetch-quest and the-api-has-opinions at release (the capture held no proof); fetch-quest went unattributed; the-api-has-opinions was flagged by the web autopilot once a changed solve.py won itday 25 web: partial, single at 2:23 (unattributed). the matcher left proof-1 unanswered on shell-company, reset-expectations and resolver-revolt at release (the capture held no proof); reset-expectations and shell-company went unattributed; resolver-revolt was flagged by the web autopilot once a changed solve.py won itday 26 web: long, single at 17:45 (unattributed). the matcher left proof-1 unanswered on trust-the-webhook and serialized-offender at release (the capture held no proof); trust-the-webhook went unattributed; serialized-offender was flagged by the web autopilot once a changed solve.py won itday 27 web: long, single at 25:56 (unattributed). the matcher left proof-1 unanswered on socket-to-me, forwarded-misgivings and double-spent at release (the capture held no proof); socket-to-me went unattributed; forwarded-misgivings was flagged by the web autopilot once a changed solve.py won it; double-spent was flagged by a web workerday 28 web: cold start, single at 3:25 (web autopilot). the matcher left proof-1 unanswered on fail-open-friday and if-a-breach-falls-in-a-log at release (the capture held no proof); fail-open-friday and if-a-breach-falls-in-a-log were flagged by the web autopilot once a changed solve.py won themday 29 web: long, single at 31:39 (web worker). the matcher left proof-1 unanswered on trust-the-build and internal-affairs at release (the capture held no proof; the solver set gave no answer); trust-the-build was flagged by the web autopilot once a changed solve.py won it; internal-affairs was flagged by a web workerday 30 web: partial, single at 1:04 (web autopilot). the matcher left proof-1 unanswered on pass-the-identity and queue-the-confusion at release (a proofless capture, no crash); pass-the-identity and queue-the-confusion were flagged by the web autopilot once a changed solve.py won themday 31 web: cold start, single at 9:22 (web autopilot). the matcher left proof-1 unanswered on framework-roulette and full-stack-incident at release (a proofless capture, no crash); full-stack-incident and framework-roulette were flagged by the web autopilot once a changed solve.py won them
  • covered at release
  • partial, under 3 min
  • cold start, 3 to 15 min
  • long, over 15 min
  • no release pass or unknown

hover a cell for the time, who took the flag and what was missing. grey on days 1 to 4: autopilot's release pass started on day 5. the grey cell on day 5 is binary, where the pass failed on a missing python module before it could post a gaps row. the web row starts on day 19, when the lane joined the board; its release pass is the web autopilot slot and its misses went to a racer.

the web row covers days 19 to 31, 13 days: 3 partial, 5 cold start, 4 long, 1 no release pass. its release pass is the web autopilot slot; a miss went to a racer and, failing that, a reasoner. its times are the day's last web single on the public board.

what the release pass missed

the first chart counts the field families in each day's gaps rows; the second times the singles those rows were about.

field families the release pass could not answer, per day
05101520day 1: no release pass yet1day 2: no release pass yetday 3: no release pass yet3day 4: no release pass yetday 5: 11 families (mobile 1, script 10)5day 6: 61 families (metadata 1, script 58, binary 2) (off the scale)day 7: 33 families (metadata 1, script 32) (off the scale)7day 8: 34 families (metadata 2, script 28, binary 4) (off the scale)day 9: 2 families (metadata 1, binary 1)9day 10: 2 families (metadata 1, binary 1)day 11: none, every single covered11day 12: 1 families (metadata 1)day 13: 1 families (metadata 1)13day 14: 1 families (metadata 1)day 15: 2 families (metadata 1, binary 1)15day 16: none, every single coveredday 17: 14 families (warmups 1, metadata 1, mobile 1, binary 11)17day 18: 1 families (metadata 1)day 19: 2 families (warmups 1, binary 1)19day 20: 4 families (metadata 1, mobile 1, binary 1, web 1)day 21: 3 families (metadata 1, binary 1, web 1)21day 22: 1 families (web 1)day 23: 2 families (metadata 1, web 1)23day 24: 2 families (warmups 1, web 1)day 25: 2 families (metadata 1, web 1)25day 26: 4 families (metadata 1, mobile 1, binary 1, web 1)day 27: 1 families (web 1)27day 28: 19 families (metadata 1, mobile 1, binary 16, web 1)day 29: 2 families (mobile 1, web 1)29day 30: 3 families (metadata 1, binary 1, web 1)day 31: 3 families (metadata 1, binary 1, web 1)31

a family is the answer-id prefix a gaps row groups its fields by (`d12-dotnet`, `answer`, `python-pyinstaller`). the scale stops at 20; a dashed bar is a day beyond it, where the script lane's descriptor families ran to dozens (hover for the count by lane).

release to the single's flag, when the release pass left a gap
0:001:152:303:455:00day 1: no release pass yet1day 2: no release pass yetday 3: no release pass yet3day 4: no release pass yetday 5: script 3:29 (cold start), mobile 1:16 (partial), binary 0:28 (unknown)5day 6: binary 1:40 (partial), script 1:27 (partial), metadata 1:04 (partial)day 7: script 2:25 (partial), metadata 1:05 (partial)7day 8: script 2:07 (partial), binary 1:54 (partial), metadata 1:06 (partial)day 9: binary 1:03 (partial), metadata 0:26 (partial)9day 10: metadata 1:08 (partial), binary 1:02 (partial)day 11: every single covered at release11day 12: metadata 1:03 (partial)day 13: metadata 1:26 (partial)13day 14: metadata 0:18 (partial)day 15: binary 1:02 (partial), metadata 0:23 (partial)15day 16: every single covered at releaseday 17: binary 1h41m (long), warmups 0:42 (partial), mobile 0:19 (partial), metadata 0:18 (partial) (off the scale)17day 18: metadata 0:18 (partial)day 19: binary 0:54 (partial), warmups 0:26 (partial)19day 20: web 6:02 (cold start), binary 0:42 (partial), metadata 0:38 (partial), mobile 0:02 (partial) (off the scale)day 21: web 23:04 (long), metadata 0:30 (partial), binary 0:22 (partial) (off the scale)21day 22: web 8:08 (cold start) (off the scale)day 23: web 4:40 (cold start), metadata 0:15 (partial)23day 24: web 1:54 (partial), warmups 0:21 (partial)day 25: web 2:23 (partial), metadata 0:18 (partial)25day 26: web 17:45 (long), metadata 1:28 (partial), binary 1:02 (partial), mobile 0:29 (partial) (off the scale)day 27: web 25:56 (long) (off the scale)27day 28: mobile 14:12 (cold start), binary 13:37 (cold start), metadata 8:16 (cold start), web 3:25 (cold start) (off the scale)day 29: web 31:39 (long), mobile 0:17 (partial) (off the scale)29day 30: binary 3:58 (cold start), web 1:04 (partial), metadata 0:56 (partial)day 31: web 9:22 (cold start), metadata 1:11 (partial), binary 1:08 (partial) (off the scale)31

one bar per day: the slowest single that autopilot did not answer at release (every such lane is in the hover). the scale stops at five minutes; a dashed bar is a day beyond it. no bar: every single was covered, or the day ran before the release pass.

day by day

each table is one day. at release is the cell colour above; what was missing is read from autopilot's gaps row (the field count, the families and the reason its crash gives); closed by is the role on the single's flag; in is release to that flag; what landed that night links the tooling sections written at the end of that day, which the next morning's solver carried. the warmups row also names the day's codec and when it entered the decoder.

day 01

no release pass yet: an agent took every single. the board opened at 11:28, so these times count from 09:00 and include the late start.

lane at release what was missing closed by in what landed that night
warmups no release pass codec b64: in the first commit (day 4), where the git history starts unattributed 2h30m -
metadata no release pass - unattributed 2h30m -
mobile no release pass - unattributed 2h48m -
script no release pass - unattributed 2h45m -
binary no release pass - unattributed 2h43m -

day 02

no release pass yet: an agent took every single.

lane at release what was missing closed by in what landed that night
warmups no release pass codec rot13: in the first commit (day 4), where the git history starts helper 10:25 -
metadata no release pass - parent 9:06 -
mobile no release pass - lane fixer 9:56 -
script no release pass - helper 12:35 -
binary no release pass - parent 7:10 -

day 03

no release pass yet: an agent took every single.

lane at release what was missing closed by in what landed that night
warmups no release pass codec hex: in the first commit (day 4), where the git history starts helper 0:26 -
metadata no release pass - parent 0:28 prepared format readers
mobile no release pass - helper 0:30 -
script no release pass - parent 7:18 -
binary no release pass - parent 9:37 -

day 04

no release pass yet: an agent took every single. autopilot first ran at 11:44:40, once every single was flagged.

lane at release what was missing closed by in what landed that night
warmups no release pass codec bin: in the decoder since the first commit (day 4) helper 0:58 more representations
metadata no release pass - helper 1:12 jpeg frame dimensions, image dimensions in any common format
mobile no release pass - helper 0:49 keyed codecs, rc4, vigenere and compressed payloads
script no release pass - helper 3:27 vbscript and language-agnostic material, control flow in a new language
binary no release pass - helper 2:31 field producers instead of day analyzers, build facts

day 05

2 of 5 singles covered at release; the pass left a gap in mobile, script and binary.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec dec: in the decoder since the first commit (day 4) autopilot 0:02 ranked xor candidates
metadata covered nothing: autopilot answered every field at release autopilot 0:02 day 5 additions
mobile partial 1 field (answer): no decoding path found helper 1:16 day 5 additions
script cold start 126 fields in 10 families (python): ValueError: no Material_ const in vbscript; ValueError: vbscript material does not decode helper 3:29 finding the descriptor in a language we have not seen
binary unknown no gaps row: autopilot's pass on it failed before its dry run (ModuleNotFoundError: No module named 'pefile') helper 0:28 demangling, the program crate, mangling scheme

the first crash in each gaps row, as autopilot posted it:

day 06

2 of 5 singles covered at release; the pass left a gap in metadata, script and binary.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec oct: in the decoder since the first commit (day 4) autopilot 0:01 -
metadata partial 1 field (d06-pdf): no reader for d06-pdf-001-pdf-header-version helper 1:04 splitting an unknown field, a last resort for unknown properties
mobile covered nothing: autopilot answered every field at release autopilot 0:02 -
script partial 369 fields in 58 families (bash, batch, collection, impact, jscript, python, …): no crash; 5 solver answers rejected, e.g. decode_configuration,return_configurati… parent 1:27 day 6 additions
binary partial 8 fields (d06-analysis, d06-artifact): a KeyError on 06 helper 1:40 tagged members, whatever their format, constant pools

the first crash in each gaps row, as autopilot posted it:

day 07

3 of 5 singles covered at release; the pass left a gap in metadata and script.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec url: in the decoder since the first commit (day 4) autopilot 0:05 numbered bases not used yet, scoring line combinations once
metadata partial 1 field (d07-zip): no reader for zip-eocd-total-entries helper 1:05 raw header fields, matched by name, zip census, nested archives as members
mobile covered nothing: autopilot answered every field at release autopilot 0:04 keyed steps on raw text, codec names from other lanes
script partial 255 fields in 32 families (bash, nodejs, python, resource, vbscript): a KeyError on persistence-adapter, persistence-attack, persistence-campaign parent 2:25 placing a field by the file its question names, roles from the question, history as a fallback, labels read both ways, the whole label space as a test, joining split literals outside the strings, compiled carriers
binary covered nothing: autopilot answered every field at release autopilot 0:05 naming the format from the bytes, dex constants, mach-o, webassembly, failures are errors

machinery that night, solver pipeline: nested archives per set.

the first crash in each gaps row, as autopilot posted it:

day 08

2 of 5 singles covered at release; the pass left a gap in metadata, script and binary.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec escapes: in the decoder since the first commit (day 4) autopilot 0:04 ties and budgets in the line decoder, the whole message, scored by letter pairs, more representations
metadata partial 2 fields (d07-zip, evidence-note): no reader for zip-central-directory-member-names helper 1:06 central directory names, a guess must cover every word, bounded decompression
mobile covered nothing: autopilot answered every field at release autopilot 0:04 more codec names, list data
script partial 313 fields in 28 families (host, polyglot, python, resource, vbscript): no answer for host-php-webshell-php-obfuscation, polyglot-php-benign-utility-obfuscation, polyglot-php-launcher-obfuscation helper 2:07 analysis headers, stated values win
binary partial 11 fields in 4 families (d02, d06, d07, d08): no producer for compression_methods, has_cpp_typeinfo, has_cpp_vtable helper 1:54 hcfg in a base64 literal, answers by kind, fields for old tags, faster on large members

machinery that night, solver pipeline: reading the answer format, field names without set prefixes.

the first crash in each gaps row, as autopilot posted it:

day 09

3 of 5 singles covered at release; the pass left a gap in metadata and binary.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec jsonstr: jsonstr entered the decoder that day at 09:45, after the single's flag autopilot 0:04 quoted lines, the literal family, more representations, caesar shifts and two-step repair, keyed lines, failure reports and confidence
metadata partial 1 field (d09-pe): no reader for pe-coff-machine-u16-hex helper 0:26 header fields by structured name, loud failures
mobile covered nothing: autopilot answered every field at release autopilot 0:04 any registry transform is a codec name, every reading of the data and the key, keys from the known prefix, tolerant answer extraction, asset and field guards
script covered nothing: autopilot answered every field at release autopilot 0:05 relations by field grammar, a stage graph from descriptor claims, an impossible family tree, ranked candidate labels, tables that follow their files, rows need evidence
binary partial 1 field (d09-analysis): no producer for hcfg_offset helper 1:03 one carrier search, the offset, old tags answer from their format or fail, faster across sets, stripped nim builds

machinery that night, solver pipeline: failures recorded per set, hints and low-confidence answers.

the first crash in each gaps row, as autopilot posted it:

day 10

3 of 5 singles covered at release; the pass left a gap in metadata and binary.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec b32: in the decoder since the first commit (day 4) autopilot 0:04 base32 that tolerates its own padding, base32's other alphabets, number runs and a faster failure
metadata partial 1 field (d09-pe): no reader for pe-address-of-entry-point-rva-hex helper 1:08 shorter structured names, the pe version resource, executable labels and zip provenance
mobile covered nothing: autopilot answered every field at release autopilot 0:04 split, again and changed, manifest and network security config
script covered nothing: autopilot answered every field at release autopilot 0:05 wmi subscriptions, carriers
binary partial 1 field (d10-analysis): no producer for has_hcfg_section helper 1:02 section-table questions, runtime markers that survive strip, jvm, kotlin and swift structure, the dynamic trace

machinery that night, solver pipeline: a confirmed candidate first.

the first crash in each gaps row, as autopilot posted it:

day 11

all 5 singles covered at release.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec b64url: in the decoder since the first commit (day 4) autopilot 0:01 the rest of the base family
metadata covered nothing: autopilot answered every field at release autopilot 0:01 the rest of the pe, comparing a pe's names, archives
mobile covered nothing: autopilot answered every field at release autopilot 0:01 url-safe base64 under a keyed step
script covered nothing: autopilot answered every field at release autopilot 0:02 escaped bash material, one more known prefix
binary covered nothing: autopilot answered every field at release autopilot 0:02 where an offset sits, fewer external tool runs

day 12

4 of 5 singles covered at release; the pass left a gap in metadata.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec reverse: in the decoder since the first commit (day 4) autopilot 0:02 whole-text transforms, a search does not outrank a codec
metadata partial 1 field (d12-dotnet): no reader for dotnet-mvid-guid helper 1:03 .net identity by name
mobile covered nothing: autopilot answered every field at release autopilot 0:02 app fields from the other lanes, two more codec names
script covered nothing: autopilot answered every field at release autopilot 0:03 fields from the other lanes
binary covered nothing: autopilot answered every field at release autopilot 0:06 fields from the other lanes, .net columns by name

the first crash in each gaps row, as autopilot posted it:

day 13

4 of 5 singles covered at release; the pass left a gap in metadata.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec atbash: in the decoder since the first commit (day 4) autopilot 0:02 classical substitutions after atbash
metadata partial 1 field (d13-msi): no reader for msi-product-code-guid parent 1:26 windows installer databases, fields across members
mobile covered nothing: autopilot answered every field at release autopilot 0:02 aes codecs
script covered nothing: autopilot answered every field at release autopilot 0:03 batch material shifted by a constant, the expansion label read from the statement
binary covered nothing: autopilot answered every field at release autopilot 0:05 -

the first crash in each gaps row, as autopilot posted it:

day 14

4 of 5 singles covered at release; the pass left a gap in metadata.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec rot19: in the decoder since the first commit (day 4) autopilot 0:03 line k in day k's codec, non-ascii and esoteric encodings
metadata partial 1 field (d14-lnk): no reader for lnk-target-path-windows helper 0:18 one-record readers, mapped by the property's words
mobile covered nothing: autopilot answered every field at release autopilot 0:02 transpositions, substitutions and a key the app holds
script covered nothing: autopilot answered every field at release autopilot 0:04 more material transforms, and a header without its comment
binary covered nothing: autopilot answered every field at release autopilot 0:05 -

the first crash in each gaps row, as autopilot posted it:

day 15

3 of 5 singles covered at release; the pass left a gap in metadata and binary.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec rot47: in the decoder since the first commit (day 4) autopilot 0:02 -
metadata partial 1 field (d15-ooxml): no reader for ooxml-external-relationship-targets lane fixer 0:23 ooxml relationships
mobile covered nothing: autopilot answered every field at release autopilot 0:02 -
script covered nothing: autopilot answered every field at release autopilot 0:03 day 15 additions
binary partial 8 fields (d15-analysis): no producer for stripped_architecture, stripped_config_campaign, stripped_config_domain lane fixer 1:02 a role in the field name, functions of a c build

machinery that night, cross-lane fields: day 15 additions.

the first crash in each gaps row, as autopilot posted it:

day 16

all 5 singles covered at release.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec morse>hex: a chain of morse and hex, each in the decoder since the first commit (day 4); the chain's row was added on day 16 autopilot 0:03 two more schedule rows, and rows that chain, dot and dash name morse, a short new line, ranked by slug words, the failure report leads with what the table did not read
metadata covered nothing: autopilot answered every field at release autopilot 0:05 mail threads
mobile covered nothing: autopilot answered every field at release autopilot 0:03 a recorded stand-in
script covered nothing: autopilot answered every field at release autopilot 0:05 -
binary covered nothing: autopilot answered every field at release autopilot 0:04 sections, instructions, decoded, a recorded stand-in

machinery that night, solver pipeline: a run saves the answers that were accepted; replay and analysis: the table check, day 16 promoted.

day 17

1 of 5 singles covered at release; the pass left a gap in warmups, metadata, mobile and binary.

lane at release what was missing closed by in what landed that night
warmups partial 1 field (message): the dry run ran past its 15 s cap; codec baconhex: baconhex entered the decoder that day at 11:14, after the single's flag lane fixer 0:42 bacon groups as hex nibbles
metadata partial 1 field (d17-audio): no reader for media-primary-audio-codec-name lane fixer 0:18 media streams by ffprobe's own keys
mobile partial 1 field (answer): no decoding path found lane fixer 0:19 one character per index
script covered nothing: autopilot answered every field at release autopilot 0:03 -
binary long 12 fields in 11 families (d02, d04, d05, d06, d07, d08, …): no producer for access_flags, has_coroutine_runtime, has_go_runtime helper, after 1 re-run of a saved fix 1h41m the next keys of each record, the storage form of the configuration blob, c++ names on a pe build, runtime markers and a class's access word

the first crash in each gaps row, as autopilot posted it:

day 18

4 of 5 singles covered at release; the pass left a gap in metadata.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec railfence3: railfence3 entered the decoder that day at 09:45, after the single's flag autopilot 0:01 a fixed rail count for the line schedule, a keyless columnar transposition
metadata partial 1 field (d18-video): KeyError: "media-primary-video-dimensions: the video stream has no 'dimensions'" lane fixer 0:18 composites, near misses and the keys a stream has, video containers in the sniff, composites of any reader's keys
mobile covered nothing: autopilot answered every field at release autopilot 0:01 chacha20, triple des and blowfish beside aes
script covered nothing: autopilot answered every field at release autopilot 0:03 -
binary covered nothing: autopilot answered every field at release autopilot 0:03 -

the first crash in each gaps row, as autopilot posted it:

day 19

3 of 6 singles covered at release; the pass left a gap in warmups, binary and web.

lane at release what was missing closed by in what landed that night
warmups partial 1 field (message): no decoding path found; codec utf16lehex: utf16lehex entered the decoder that day at 12:43, after the single's flag parent 0:26 wide text under every step, the nineteenth schedule row and wide-text hints
metadata covered nothing: autopilot answered every field at release autopilot 0:02 a header field named without a format word
mobile covered nothing: autopilot answered every field at release autopilot 0:02 -
script covered nothing: autopilot answered every field at release autopilot 0:05 an analysis header inside a string literal
binary partial 8 fields (d19-analysis): no producer for overlay_offset, overlay_size; no valid hcfg blob in any member autopilot, re-running a lane fixer's save 0:54 finding the container boundary, proving the unpacking step, feeding the existing record parser
web no release pass - web worker 2:32 -

the first crash in each gaps row, as autopilot posted it:

day 20

2 of 6 singles covered at release; the pass left a gap in metadata, mobile, binary and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec utf16behex: utf16behex entered the decoder that day at 10:05, after the single's flag autopilot 0:01 the big-endian row
metadata partial 1 field (d20-macho): no reader for macho-cpu-type-set lane fixer 0:38 a field over every record: 0 values, the bytes pick the struct
mobile partial 1 field (answer): no decoding path found autopilot 0:02 data that names a member, the normal reader first
script covered nothing: autopilot answered every field at release autopilot 0:02 -
binary partial 1 field (d20-analysis): no producer for hcfg_resource_size autopilot, re-running a lane fixer's save 0:42 a header with scratch after the data, the question's words for the layout
web cold start the matcher left proof-1 unanswered on lfi at release (no request or static source rule for proof-1); administratively-speaking went unattributed without an autopilot pass; lfi was flagged by the parent parent 6:02 -

the first crash in each gaps row, as autopilot posted it:

day 21

3 of 6 singles covered at release; the pass left a gap in metadata, binary and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec b64inflate: in the decoder since the first commit (day 4) autopilot 0:02 a compressed line
metadata partial 1 field (d21-apk): no reader for apk-package-name lane fixer, after 1 re-run of a saved fix 0:30 an android package as a record
mobile covered nothing: autopilot answered every field at release autopilot 0:01 every manifest and signer scalar by name
script covered nothing: autopilot answered every field at release autopilot 0:02 -
binary partial 6 fields (d21-analysis): no producer for loader_machine, plugin_config_campaign, plugin_config_domain autopilot, re-running a lane fixer's save 0:22 role pairs across extensions
web long the matcher left proof-1 unanswered on session-impression, stored-xss and reflect-upon-this at release (the capture held no proof); all three singles went unattributed unattributed 23:04 -

machinery that night, cross-lane fields: the package name across categories.

the first crash in each gaps row, as autopilot posted it:

day 22

5 of 6 singles covered at release; the pass left a gap in web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec b64inflate: in the decoder since the first commit (day 4) autopilot 0:02 a zlib line, and the rest of the wrapper family, a slug that runs past its shape, hint words by stem
metadata covered nothing: autopilot answered every field at release autopilot 0:02 the jar record family
mobile covered nothing: autopilot answered every field at release autopilot 0:02 -
script covered nothing: autopilot answered every field at release autopilot 0:04 -
binary covered nothing: autopilot answered every field at release autopilot 0:08 -
web cold start the matcher left proof-1 unanswered on forged-with-care and uploading-expectations at release (the capture held no proof); uploading-expectations went unattributed; forged-with-care was flagged by a web worker unattributed 8:08 -

machinery that night, replay and analysis: a rejected run says what changed.

the first crash in each gaps row, as autopilot posted it:

day 23

4 of 6 singles covered at release; the pass left a gap in metadata and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec a85: in the decoder since the first commit (day 4) autopilot 0:01 ascii85 as the twenty-third row, a short new line read as written
metadata partial 1 field (d23-sqlite): no crash; the solver's 12288 was rejected autopilot, re-running a lane fixer's save 0:15 a field id splits at its stem first, the sqlite record, every key by its words, a database embedded past offset 0, member-name lists in two orders
mobile covered nothing: autopilot answered every field at release autopilot 0:02 a dalvik interpreter for computed codes, native pieces under an emulator, assets, resources and a loaded dex, the dex view in triage
script covered nothing: autopilot answered every field at release autopilot 0:03 a layered analysis header, specimens named by a random stem, composite answers, batch's expansion form as code
binary covered nothing: autopilot answered every field at release autopilot 0:04 -
web cold start the matcher left proof-1 unanswered on ask-again-later, template-temper and posted-too-much at release (the capture held no proof); all three singles went unattributed unattributed 4:40 -

machinery that night, solver pipeline: a digit width the question states; replay and analysis: a saved web run belongs to its release day.

the first crash in each gaps row, as autopilot posted it:

day 24

4 of 6 singles covered at release; the pass left a gap in warmups and web.

lane at release what was missing closed by in what landed that night
warmups partial 1 field (message): no decoding path found; codec netstring: netstring entered the decoder that day at 09:32, after the single's flag lane fixer 0:21 netstrings as the twenty-fourth row, five more framings beside it, a shape gate before each new reading
metadata covered nothing: autopilot answered every field at release autopilot 0:02 a member named by the leading tokens of its stem, the event-log record, read as a whole log
mobile covered nothing: autopilot answered every field at release autopilot 0:02 -
script covered nothing: autopilot answered every field at release autopilot 0:04 -
binary covered nothing: autopilot answered every field at release autopilot 0:07 -
web partial the matcher left proof-1 unanswered on fetch-quest and the-api-has-opinions at release (the capture held no proof); fetch-quest went unattributed; the-api-has-opinions was flagged by the web autopilot once a changed solve.py won it unattributed 1:54 -

machinery that night, solver pipeline: a crash report says what each field asked; replay and analysis: a passing regression run, keyed on the code it checked, the next-day model learns the new row.

the first crash in each gaps row, as autopilot posted it:

day 25

4 of 6 singles covered at release; the pass left a gap in metadata and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec uniesc: in the decoder since day 8 autopilot 0:01 code points as the twenty-fifth row
metadata partial 1 field (d25-registry): no reader for registry-hive-root-name autopilot, re-running a lane fixer's save 0:18 the hive record's whole-tree keys, one word over: a property's synonym
mobile covered nothing: autopilot answered every field at release autopilot 0:02 -
script covered nothing: autopilot answered every field at release autopilot 0:04 -
binary covered nothing: autopilot answered every field at release autopilot 0:06 -
web partial the matcher left proof-1 unanswered on shell-company, reset-expectations and resolver-revolt at release (the capture held no proof); reset-expectations and shell-company went unattributed; resolver-revolt was flagged by the web autopilot once a changed solve.py won it unattributed 2:23 -

machinery that night, cross-lane fields: readings one word over.

the first crash in each gaps row, as autopilot posted it:

day 26

2 of 6 singles covered at release; the pass left a gap in metadata, mobile, binary and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec brainfuck: in the decoder since day 14 autopilot 0:02 brainfuck as the twenty-sixth row
metadata partial 1 field (d26-prefetch): no reader for prefetch-executable-name autopilot, re-running a lane fixer's save 1:28 compressed prefetch: lzxpress huffman in python, the prefetch record
mobile partial 1 field (answer): the split parts did not join parent 0:29 a bundle of split apks is one app, 0 and a 1 over a split, a split puzzle's missing part, a missing record field names its builder
script covered nothing: autopilot answered every field at release autopilot 0:03 -
binary partial 2 fields (d26-analysis): no producer for crash_fault_method, crash_frames_match_methods autopilot, re-running a lane fixer's save 1:02 a jar's crash dump
web long the matcher left proof-1 unanswered on trust-the-webhook and serialized-offender at release (the capture held no proof); trust-the-webhook went unattributed; serialized-offender was flagged by the web autopilot once a changed solve.py won it unattributed 17:45 -

the first crash in each gaps row, as autopilot posted it:

day 27

5 of 6 singles covered at release; the pass left a gap in web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec ook: in the decoder since day 14 autopilot 0:02 ook! as the twenty-seventh row
metadata covered nothing: autopilot answered every field at release autopilot 0:03 -
mobile covered nothing: autopilot answered every field at release autopilot 0:02 -
script covered nothing: autopilot answered every field at release autopilot 0:04 -
binary covered nothing: autopilot answered every field at release autopilot 0:09 -
web long the matcher left proof-1 unanswered on socket-to-me, forwarded-misgivings and double-spent at release (the capture held no proof); socket-to-me went unattributed; forwarded-misgivings was flagged by the web autopilot once a changed solve.py won it; double-spent was flagged by a web worker unattributed 25:56 -

the first crash in each gaps row, as autopilot posted it:

day 28

2 of 6 singles covered at release; the pass left a gap in metadata, mobile, binary and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec reverse>hex: a chain of reverse and hex, each in the decoder since the first commit (day 4); the chain's row was added on day 28 autopilot 0:03 reversed hex as the twenty-eighth row, no beam after a confident slug, a budget per blob, seven more readings, a distinct-letter guard on substitutions
metadata cold start 1 field (d28-disk): no reader for disk-partition-start-lba-sequence lane fixer 8:16 a disk record for partitioned images, the partition an offset in the question names, filesystem types and container headers, header-first lookups, the winzip-aes key through hashlib
mobile cold start 1 field (answer): no decoding path found parent 14:12 the packet codec: a frame layout chosen by its checksums, an answer-shape scan, a framework layer for the dalvik interpreter
script covered nothing: autopilot answered every field at release autopilot 0:04 bare runs as a last-resort carrier
binary cold start 32 fields in 16 families (d01, d03, d05, d06, d09, d10, …): no producer for class_name, cli_runtime, conditional_branch_opcode_count parent 13:37 questions read as recipes, readers that dispatch on the member's format, the family and its false friend
web cold start the matcher left proof-1 unanswered on fail-open-friday and if-a-breach-falls-in-a-log at release (the capture held no proof); fail-open-friday and if-a-breach-falls-in-a-log were flagged by the web autopilot once a changed solve.py won them web autopilot 3:25 -

machinery that night, solver pipeline: a day's field rules carried into the next day's solver; replay and analysis: the end-of-day regression gate as one command, the next-day model learns row 28.

the first crash in each gaps row, as autopilot posted it:

day 29

4 of 6 singles covered at release; the pass left a gap in mobile and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec rot13>b64: a chain of rot13 and b64, each in the decoder since the first commit (day 4); the chain's row was added on day 29 autopilot 0:02 rot13 over base64 as the twenty-ninth row
metadata covered nothing: autopilot answered every field at release autopilot 0:03 six more record readers
mobile partial 1 field (answer): no decoding path found lane fixer 0:17 the table codec: a byte table keyed by its index
script covered nothing: autopilot answered every field at release autopilot 0:04 -
binary covered nothing: autopilot answered every field at release autopilot 0:07 -
web long the matcher left proof-1 unanswered on trust-the-build and internal-affairs at release (the capture held no proof; the solver set gave no answer); trust-the-build was flagged by the web autopilot once a changed solve.py won it; internal-affairs was flagged by a web worker web worker 31:39 -

machinery that night, solver pipeline: a field rule that raises falls back to the lane reader.

the first crash in each gaps row, as autopilot posted it:

day 30

3 of 6 singles covered at release; the pass left a gap in metadata, binary and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec url>url: a chain of url and url, each in the decoder since the first commit (day 4); the chain's row was added on day 30 autopilot 0:02 url twice as the thirtieth row, a chain of the day table's codecs, a line nothing reads goes out unsure, a short line read by a search is never confident
metadata partial 1 field (d30-cluster): no reader for relation-identical-sha256-groups autopilot, after 1 re-run of a saved fix 0:56 the relation family, semantic keys, near keys and crops, capture hosts and one tool budget, memos keyed by digest
mobile covered nothing: autopilot answered every field at release autopilot 0:03 the layered codec: reversed bytes under a walking key, a bounded layer peel, two keyed layers and an unnamed block step, an ambiguous stack and the unstripped root
script covered nothing: autopilot answered every field at release autopilot 0:04 a guessed label goes out unsure, three readers of the descriptor graph
binary cold start 8 fields (d30-analysis): no valid hcfg blob in any member; an ungraded binary format, neko_bytecode autopilot, re-running a lane fixer's save (1 more re-run before it) 3:58 a configuration record under a renamed magic, every carrier stage without a magic, producers for techniques never asked by name, the call graph inside a timed set
web partial the matcher left proof-1 unanswered on pass-the-identity and queue-the-confusion at release (a proofless capture, no crash); pass-the-identity and queue-the-confusion were flagged by the web autopilot once a changed solve.py won them web autopilot 1:04 -

machinery that night, cross-lane fields: a miss report that always ends in a draft, other lanes' readers as relation keys; solver pipeline: a learned candidate position needs the same count, a far edge drops the pin, the dashboard's board and second panel, the shell commands run in; replay and analysis: families named in advance, the forward matrix, the next-day model learns row 30, a base with no timed tier is no lane base.

the first crash in each gaps row, as autopilot posted it:

day 31

3 of 6 singles covered at release; the pass left a gap in metadata, binary and web.

lane at release what was missing closed by in what landed that night
warmups covered nothing: autopilot answered every field at release; codec b64inflate>b32: a chain of b64inflate and b32, each in the decoder since the first commit (day 4); the chain's row was added on day 31 autopilot 0:02 -
metadata partial 1 field (d31-case): no reader for graph-execution-chain-artifact-ids lane fixer 1:11 -
mobile covered nothing: autopilot answered every field at release autopilot 0:03 -
script covered nothing: autopilot answered every field at release autopilot 0:03 -
binary partial 10 fields (d31-analysis): no producer for config_campaign, config_domain, config_mutex lane fixer, after 1 re-run of a saved fix 1:08 -
web cold start the matcher left proof-1 unanswered on framework-roulette and full-stack-incident at release (a proofless capture, no crash); full-stack-incident and framework-roulette were flagged by the web autopilot once a changed solve.py won them web autopilot 9:22 -

the first crash in each gaps row, as autopilot posted it:

what kept coming back

a family that one lane's release pass missed on two or more days:

lane families days
mobile answer 5, 17, 20, 26, 28, 29
binary d06-analysis 6, 8, 17, 28
script python-current, python-cxfreeze, python-embedded, … (10 families) 5, 6, 7, 8
binary d09-analysis 9, 17, 28
script resource-attack, resource-mutex, resource-scenario, … (14 families) 6, 7, 8
warmups message 17, 19, 24
binary d02-analysis, d07-analysis, d08-analysis 8, 17
binary d05-analysis, d11-analysis, d12-analysis, … (4 families) 17, 28
binary d10-analysis 10, 28
binary d19-analysis 19, 28
binary d26-analysis 26, 28
metadata d07-zip 7, 8
metadata d09-pe 9, 10

metadata left a gap on 19 of 27 raced days, the most of any lane, and no metadata gaps row named more than 2 fields. on 17 of those days the crash named a property no reader had.

mobile has one field, answer, and it was the gap on 6 days (5, 17, 20, 26, 28 and 29). on 5 of them the decoder found no decoding path through the day's puzzle (day 26: the split parts did not join).

binary's bundles kept re-asking older day tags with fields the lane had no producer for: d02-analysis, d05-analysis, d06-analysis, d07-analysis, d08-analysis, d09-analysis, d10-analysis, d11-analysis, d12-analysis, d13-analysis, d19-analysis and d26-analysis each came back on a later day, and one binary gaps row named up to 16 families.

script's gaps all fall on days 5 to 8, each row naming 10 to 58 descriptor families; from day 9 the script single was covered at release on 23 of the 23 days.

warmups left a gap on days 17, 19 and 24; on 3 of them the day's codec (baconhex, utf16lehex, netstring) entered the decoder that day, after the single's flag. on days 9, 18 and 20 the codec also entered the decoder that day, but autopilot had already read the single at release with the decoder it had.

where these numbers come from

days 1 to 31. the misses are autopilot's own intel rows: a release-pass gaps row per single it could not answer (from day 5), the is yours rows that handed a field to the lane fixer and the re-ran the changed solve.py rows that followed a save, from the archived intel of each day and the live intel of day 31, deduplicated by timestamp and text. each single's time and the role that took it come from the flag ledger (the grader's timestamp from day 9, the local one before), and match the race clock on every lane and day. what landed each night is every ## day N additions section of the tool pages. the warmups codec is the decoder's day table, each step dated by the first commit that added its name to the decoder; the history starts at the import on day 4, so a codec from that commit may be older. the one unknown cell reads the day's autopilot log. tools/after-action/telemetry/coverage.py writes the numbers and blog/site/coverage.py this page; the web lane and the day 30 static challenge are left out.

{generated by a clanker}