Kjell Tore Guttormsen daa12b3bbb test(llm-security): narrow av-surface probe (b) to a command shape ... The S0 rule fired on 'http' anywhere in decoded text and flagged an SVG data URI (xmlns) in the hyperframes-like skill-scan fixture - markup, not a command. PM decision for S2: the decoded text must start with a command word or contain a pipe into a shell or $(. Measured before any blob was removed: 8 blobs/7 files -> 8 blobs/6 files; SVG out, one split-payload fragment in security-assessment.md newly caught, all 7 command blobs kept. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>		2026-09-22 13:17:02 +02:00
..
e2e	test(llm-security): build poisoned fixtures at test time, never on disk	2026-09-22 13:07:38 +02:00
fixtures	test(llm-security): build poisoned fixtures at test time, never on disk	2026-09-22 13:07:38 +02:00
golden	refactor(llm-security): build the SIG ruleset from vendored commons (malware-signatures 0.1.0)	2026-08-13 21:28:14 +02:00
helpers	test(llm-security): build poisoned fixtures at test time, never on disk	2026-09-22 13:07:38 +02:00
hooks	refactor(llm-security): build the secret table from vendored commons (secret-egress 0.3.0)	2026-08-13 21:10:14 +02:00
lib	fix(llm-security): compileRules coerces a non-string pattern instead of dropping the rule	2026-08-13 21:51:10 +02:00
scanners	test(llm-security): build poisoned fixtures at test time, never on disk	2026-09-22 13:07:38 +02:00
av-surface.test.mjs	test(llm-security): narrow av-surface probe (b) to a command shape	2026-09-22 13:17:02 +02:00