open/llm-security
1
0
Fork 0
Code Issues Releases 1 Activity

Squashed 'scanners/commons/' changes from 0ffee85..532d70d

Browse source 
532d70d release(0.2.0): the contract gained a MUST, so the tag has to move
946f51d fix(active-content,conformance): cite line numbers per commit — they do not resolve at the pin
bdcb1f1 feat(conformance): ship the six active-content cases; the id space already existed
807c0d4 feat(spec): add not-applicable, so a single-runtime table stops reading as 7 defects
a1578e6 fix(conformance): record the guard's internal-surface position on _LEX_PAYLOADS
4d351d2 fix(mapping): state that three of four OWASP maps have no production consumer
f082a91 fix(lexicon,docs): retract the claim that the guard's port cites severity.mjs

git-subtree-dir: scanners/commons
git-subtree-split: 532d70d5ed2f9b23a8efad760ef490356cf52ada
This commit is contained in:
Kjell Tore Guttormsen 2026-08-11 13:02:33 +02:00
commit 9b95fc607e
20 changed files with 533 additions and 37 deletions
Download patch file Download diff file Expand all files Collapse all files

5
lexicon/injection-lexicon.json
View file

@ -1,5 +1,5 @@
{
"version": "0.5.0",
"version": "0.5.1",
"id": "injection-lexicon",
"description": "Prompt-injection pattern lexicon: the four pattern families a detector matches against untrusted text, and the severity family each belongs to. Data only. The variant building that feeds these patterns (normalisation, homoglyph folding, rot13, unicode-tag escalation) and the buried-payload heuristic are engine behaviour and are deliberately NOT described here.",
"owasp": "LLM01",
@ -894,7 +894,8 @@
"source_lines": "274-281",
"source_commit": "b0de0ca6d86ce697f39669d177c2c2654c280128",
"verified": "directly, by reading the module at the pinned commit",
"not_from": "scanners/lib/severity.mjs, which was the first file asked about and which contains no injection-family severity at all - grep for CRITICAL_PATTERNS, HIGH_PATTERNS, MEDIUM_PATTERNS or HYBRID_PATTERNS there returns nothing. The Python guard's port carries the correct value but cites that file, so a consumer following its citation would find no answer. Recorded because a wrong citation to a right value is the harder defect to notice."
"not_from": "scanners/lib/severity.mjs, which was the first file asked about and which contains no injection-family severity at all - grep for CRITICAL_PATTERNS, HIGH_PATTERNS, MEDIUM_PATTERNS or HYBRID_PATTERNS there returns nothing at the pinned commit. That part stands and was re-measured 2026-08-10.",
"retracted_2026-08-10": "This field also said: 'The Python guard's port carries the correct value but cites that file, so a consumer following its citation would find no answer.' RETRACTED - it is false, and it was never measured here. It restated an assertion received from llm-security (coord message 20260809T201048Z) as if it were a commons finding. Measured against the guard's own tree: severity.mjs has never appeared in src/llm_ingestion_guard/injection_lexicon.json at any point in that file's history (git log -S returns no commits), and at 0bf0729 - the commit conformance/manifest.json pins - the only tree-wide occurrence is docs/PLAN.md:114, which correctly attributes the report module to output.mjs + severity.mjs and says nothing about injection-family severity. The guard's only source statement for the lexicon is the note field at injection_lexicon.json:3, and it cites injection-patterns.mjs - the right file. Kept rather than deleted because this repository's stated reason for recording the original claim was that a wrong citation to a right value is the harder defect to notice, and that turned out to describe this record itself."
},
"count": 8,
"patterns": [