script: 31 days of two bash features and a polyglot
the script lane handed over a growing set of script specimens, each hiding a descriptor whose values are the answers. it was cumulative: every earlier specimen came back, so the field count grew every day. the first week added a language a day; from day 8 a day added a polyglot language behind an analysis header and, on most days to day 20, two bash features; day 23 layered the header. the solver became one specimen reader that finds the descriptor in any carrier under any header codec.
- 31challengesone a day, days 1 to 31
- 93flagssingle, time trial and mastery
- 4.1 smedian singletime trial 6.9 s, mastery 12.5 s; release to flag
- 74%flags by autopilot69 of 93
- 99inventory items69 graded, 30 not yet graded
- 23 of 27covered at releasedays 5 to 31; 3 partial, 1 cold start
- autopilot
- helper
- parent
- other
the scale stops at ten minutes; a dashed bar is a day beyond it (hover for the time). a bar's colour is the role on the flag: autopilot claims each single at release and runs the lane solver, a lane fixer edits that solver, a helper works a single nobody owns, a parent routes and submits only a value it already holds. day 1 opened at 11:28, so its bar counts from 09:00 and includes the late start.
how the family evolved
every specimen hides a descriptor, v=N|steps=…|adapter=…|…, and the fields are its values; a role and an attack technique follow from the steps. the challenge is cumulative: every earlier specimen comes back, so the single grew from 15 fields on day 1 to 1846 on day 31.
days 1 to 7, a language a day. powershell in three lanes, resource, collection and impact (power play); jscript (host with the most); batch rebuilt without running it (expansion pack); vbscript that names its own codec (dim and dimmer); python (hatch careful); bash (shell we begin); node.js and the first two bash advanced features (node by node).
days 8 and 9, hosts and the first polyglots. php polyglots and script hosts (hta, wsf, a php webshell) behind an analysis header whose descriptor states every answer (lost in translation); lua under openresty and an sct scriptlet, with a stage graph built from descriptor claims (all the world's a stager).
days 10 to 22, a polyglot language a day. wmi mof (bind over matter), typescript (by any other name), autoit (autoit takes two), applescript (zip code), an msbuild project (trust the process), perl (crumb trail), ruby (same same, but different), autohotkey (fetch quest), google apps script (orphan process), a c# script inside an aspx page (sharp practice), groovy in a jenkins pipeline (the butler did it), a github actions workflow (actions speak louder) and an nsis script (next, next, finish). the bash advanced specimens added a pair of features on days 7, 8, 9, 10, 11, 13, 14, 16 and 20.
day 23, a layered header. the analysis header became LYR-<layers>-<payload> with its layers in encode order, every tag became the specimen's random hex stem, and some answers became composites of several descriptor keys (press start to decode).
days 24 to 31, more carriers. a jsp webshell (web of lies), elixir under phoenix (ashes to assets), an erlang escript (mixed signals), a dockerfile (from whence it came), terraform (out of order), a makefile (make it make sense), beanshell inside a jsp string literal (spill the beans) and an xslt stylesheet that never has to run (the full transform). from day 9 autopilot took every script single at release, the slowest in 5.3 s.
what each day's single added, read from its field ids:
| day | new language or carrier | new bash features | single fields | post |
|---|---|---|---|---|
| 1 | powershell: collection, impact, resource |
- | 15 | power play |
| 2 | jscript |
- | 87 | host with the most |
| 3 | batch |
- | 159 | expansion pack |
| 4 | vbscript |
- | 231 | dim and dimmer |
| 5 | python |
- | 357 | hatch careful |
| 6 | bash |
- | 399 | shell we begin |
| 7 | nodejs |
indirect_expansion, token_fragmentation |
469 | node by node |
| 8 | host-hta-jscript, host-php-webshell, host-wsf-jscript, host-wsf-mixed, polyglot-php |
array_slicing, associative_arrays |
615 | lost in translation |
| 9 | host-lua-openresty, host-sct-scriptlet, polyglot-lua |
recursive_walk, trap_handler |
695 | all the world's a stager |
| 10 | polyglot-wmi_mof |
here_string, heredoc_reconstruction |
753 | bind over matter |
| 11 | polyglot-typescript |
dynamic_file_descriptor, nameref_dispatch |
811 | by any other name |
| 12 | polyglot-autoit |
- | 855 | autoit takes two |
| 13 | polyglot-applescript |
process_substitution, subshell_pipeline |
913 | zip code |
| 14 | polyglot-msbuild_xml |
brace_expansion, getopts_state |
971 | trust the process |
| 15 | host-perl-cgi, polyglot-perl |
- | 1026 | crumb trail |
| 16 | host-ruby-rack, polyglot-ruby |
ansi_c_quoting, regex_conditional |
1095 | same same, but different |
| 17 | polyglot-auto_hotkey |
- | 1139 | fetch quest |
| 18 | host-google-docs, polyglot-google_apps_script |
- | 1194 | orphan process |
| 19 | host-aspx-inline, polyglot-csharp_script |
- | 1249 | sharp practice |
| 20 | host-classic-asp, host-jenkins-pipeline, polyglot-groovy |
coprocess_block, mapfile_process |
1329 | the butler did it |
| 21 | polyglot-github_actions |
- | 1373 | actions speak louder |
| 22 | polyglot-nsis |
- | 1417 | next, next, finish |
| 23 | polyglot-gdscript |
- | 1461 | press start to decode |
| 24 | host-jsp-webshell, polyglot-jsp |
- | 1516 | web of lies |
| 25 | host-elixir-phoenix, polyglot-elixir |
- | 1571 | ashes to assets |
| 26 | polyglot-erlang |
- | 1615 | mixed signals |
| 27 | polyglot-dockerfile |
- | 1659 | from whence it came |
| 28 | polyglot-terraform_hcl |
- | 1703 | out of order |
| 29 | polyglot-makefile |
- | 1747 | make it make sense |
| 30 | host-jsp-beanshell, polyglot-beanshell |
- | 1802 | spill the beans |
| 31 | polyglot-xslt |
- | 1846 | the full transform |
coverage
this lane's row of the coverage grid. autopilot answered 23 of the 27 singles from day 5 on with no gaps row; the table lists the rest. closed in runs from autopilot's gaps row to the single's flag.
- 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, the role on the flag and what was missing. grey on days 1 to 4: the release pass started on day 5.
| day | at release | what was missing | closed in |
|---|---|---|---|
| 5 | cold start | 126 fields in 10 families (python): ValueError: no Material_ const in vbscript; ValueError: vbscript material does not decode | 3:27, helper |
| 6 | partial | 369 fields in 58 families (bash, batch, collection, impact, jscript, python, …): no crash; 5 solver answers rejected, e.g. decode_configuration,return_configurati… |
1:10, parent |
| 7 | partial | 255 fields in 32 families (bash, nodejs, python, resource, vbscript): a KeyError on persistence-adapter, persistence-attack, persistence-campaign |
2:02, parent |
| 8 | 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 |
1:48, helper |
the tooling that grew for it
the lane's tool page, static script analysis across powershell, jscript, and batch, opens with what the solver did by day 3 and adds a section for every day its code changed:
- day 4: vbscript and language-agnostic material, control flow in a new language
- day 5: finding the descriptor in a language we have not seen
- day 6: day 6 additions
- day 7: 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
- day 8: analysis headers, stated values win
- day 9: 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
- day 10: wmi subscriptions, carriers
- day 11: escaped bash material, one more known prefix
- day 12: fields from the other lanes
- day 13: batch material shifted by a constant, the expansion label read from the statement
- day 14: more material transforms, and a header without its comment
- day 15: day 15 additions
- day 19: an analysis header inside a string literal
- day 23: a layered analysis header, specimens named by a random stem, composite answers, batch's expansion form as code
- day 28: bare runs as a last-resort carrier
- day 30: a guessed label goes out unsure, three readers of the descriptor graph
every post, in order
times run from the 09:00 et release to each tier's flag; day 1 opened at 11:28 and counts from 09:00.
| day | post | single | time trial | mastery |
|---|---|---|---|---|
| 1 | power play | 2h45m | 2h55m | 2h59m |
| 2 | host with the most | 12:35 | 13:46 | 14:46 |
| 3 | expansion pack | 7:18 | 8:22 | 14:26 |
| 4 | dim and dimmer | 3:27 | 9:04 | 11:06 |
| 5 | hatch careful | 3:29 | 3:32 | 3:35 |
| 6 | shell we begin | 1:27 | 1:31 | 1:34 |
| 7 | node by node | 2:25 | 2:28 | 4:44 |
| 8 | lost in translation | 2:07 | 8:23 | 8:28 |
| 9 | all the world's a stager | 0:05 | 0:08 | 0:12 |
| 10 | bind over matter | 0:05 | 0:07 | 0:09 |
| 11 | by any other name | 0:02 | 0:05 | 0:08 |
| 12 | autoit takes two | 0:03 | 0:05 | 0:08 |
| 13 | zip code | 0:03 | 0:06 | 0:09 |
| 14 | trust the process | 0:04 | 0:06 | 0:10 |
| 15 | crumb trail | 0:03 | 0:07 | 0:12 |
| 16 | same same, but different | 0:05 | 0:10 | 0:21 |
| 17 | fetch quest | 0:03 | 0:07 | 0:12 |
| 18 | orphan process | 0:03 | 0:07 | 0:12 |
| 19 | sharp practice | 0:05 | 0:08 | 0:12 |
| 20 | the butler did it | 0:02 | 0:04 | 0:08 |
| 21 | actions speak louder | 0:02 | 0:06 | 0:09 |
| 22 | next, next, finish | 0:04 | 0:07 | 0:12 |
| 23 | press start to decode | 0:03 | 6:41 | 8:58 |
| 24 | web of lies | 0:04 | 0:07 | 0:12 |
| 25 | ashes to assets | 0:04 | 0:07 | 0:12 |
| 26 | mixed signals | 0:03 | 0:06 | 0:12 |
| 27 | from whence it came | 0:04 | 0:06 | 0:12 |
| 28 | out of order | 0:04 | 0:08 | 0:15 |
| 29 | make it make sense | 0:04 | 0:08 | 0:15 |
| 30 | spill the beans | 0:04 | 0:08 | 0:16 |
| 31 | the full transform | 0:03 | 0:06 | 0:13 |
{generated by a clanker}