/**
 * SMI-5881 — AI_DEFENCE_PATTERNS scope promotion + maxContentLength
 * enforcement tests.
 *
 * Covers:
 *  - the 4 patterns promoted 'line' -> 'both' (AD_HTML_COMMENT_VERB/NOUN,
 *    AD_NESTED_INSTRUCTION_BLOCK, AD_ZERO_WIDTH) now catch a genuinely
 *    cross-line match that was a false negative before this PR
 *  - SMI-5879 (design §3.3): the first-occurrence-only limitation this file
 *    used to pin as a known, accepted gap is now FIXED — the RC-1 two-bound
 *    global-exec loop records every distinct matching line per pattern (up
 *    to MAX_MULTILINE_LINES_PER_PATTERN), not just the first match in the
 *    whole document, so two independently-matching occurrences now both
 *    surface.
 *  - section 3.2: maxContentLength enforcement — a lower configured
 *    maxContentLength tightens the multiline-scan effective limit, and
 *    content beyond that limit produces a visible truncation finding.
 *    SMI-5879 (design §3.4) raised the underlying cap itself from 10,000 to
 *    1,000,000 code units (see MAX_CONTENT_LENGTH_FOR_REGEX's own comment) to
 *    close the 10 KB pass-1 truncation blind spot on large SKILL.md files.
 */
export {};
//# sourceMappingURL=ai-defence-multiline-promotion.test.d.ts.map