Build multi-view quote extraction

Different attacks are visible at different stripping levels.

First time here? Paste the Context Pack first so the AI understands your project - open it from the header above.

Prompt Capsule

Implement an extraction function producing three views of every command: KEEP-DOUBLE (single-quoted content stripped, double-quoted kept - because shell metacharacters like ';' and '|' still act inside double quotes in some positions and this view exposes them), FULLY-UNQUOTED (all quoted content stripped - the view for redirection detection, since quoted '<' and '>' are inert), and KEEP-QUOTE-CHARS (content stripped but the quote characters themselves preserved - the view that reveals quote-adjacency attacks like a quoted fragment abutting a '#'). Build the underlying quote-state tracker carefully: process backslash escapes BEFORE quote toggling - inside double quotes, an escaped quote is a literal character, and toggling on it desyncs the tracker for the rest of the command. Unit-test the tracker against nasty edges: escaped quotes inside quotes, adjacent empty quotes, unterminated quotes.

Paste into EmergentFull Build: complete implementation prompt with explicit requirements

Quick is short. Full Build is recommended for most steps. Strict forces real logic when the AI keeps faking output.

Actual change check

Expected after this step

A tested extraction utility returning all three views with a correct state tracker.

Should NOT happen

An existing feature broke
A button only logs to console
Data disappears after refresh
Errors fail silently with no visible state

This is what should exist before you continue. If reality does not match, do not move on.

Your notes for this step

Track what changed, failed, or needs follow-up. Notes export with the flow.

Pass the Verify Gate to complete this step

Verify before continuing

Do not move on until every check is true. The complete button stays locked until then.

Three views produced and documented with their use casesBackslash processed before quote togglingTracker edge cases covered by tests

Do not continue if…

! Checking only the parsed token list, which is precisely what the attack bypasses
! One clever regex expected to catch all quoting tricks - each family needs its own detector
! A quote tracker that toggles state on escaped quotes, desyncing permanently mid-command
! Treating the checks as done - obfuscation is an arms race, and the corpus is the memory

If the AI messes this up

Use this when:

- it only made visual forms or placeholder UI
- buttons do nothing or only log to console
- data is not persisted - state disappears on refresh
- something that worked before is now broken

Use this when the AI fakes progress or breaks the feature. It forces a real fix.

The tracker desyncs on a command with an escaped double quote. That is the classic ordering bug: handle the backslash first (consume the next character unconditionally outside single quotes) and only then consider quote toggling - re-run the edge-case suite after the fix.