llm-security/examples
Kjell Tore Guttormsen 58d2e8ad8d
fix(av-surface): no runnable base64-to-shell line on disk
PLAN § v8.1.3 tillegg a. The bash-evasion gallery's T8 comment held a
complete base64-decode-into-shell line whose 16-char blob decodes to a
home-directory deletion, under probe (b)'s 24-char floor. Probe (b2)
found two more of the same shape, fixed here too: the gallery README's
T8 row and tests/e2e/attack-chain.test.mjs (an /etc deletion). Chosen
over an exception list because the gate should have none.

The gallery and the e2e test now encode the blob at runtime from the
same fragments; the README shows a placeholder. Gallery: 10/10 blocked.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-23 11:35:50 +02:00
..
bash-evasion-gallery fix(av-surface): no runnable base64-to-shell line on disk 2026-09-23 11:35:50 +02:00
lethal-trifecta-walkthrough feat(llm-security)!: v8 Phase 3 complete - riskScoreV1, posture heuristic, docs 2026-08-09 10:25:03 +02:00
malicious-skill-demo docs(llm-security): describe the base64 exfil payloads instead of showing them 2026-09-22 13:23:24 +02:00
mcp-rug-pull feat(llm-security): add lethal-trifecta + mcp-rug-pull example contents [skip-docs] 2026-05-05 14:45:39 +02:00
poisoned-claude-md test(llm-security): store the poisoned-claude-md fixture encoded 2026-09-22 13:22:10 +02:00
pre-compact-poisoning feat(llm-security): add pre-compact-poisoning example for PreCompact hook [skip-docs] 2026-05-05 15:23:10 +02:00
prompt-injection-showcase feat(ultraplan-local): v1.6.0 — /ultraresearch-local deep research command 2026-04-08 08:58:35 +02:00
supply-chain-attack feat(llm-security)!: v8 Phase 3 complete - riskScoreV1, posture heuristic, docs 2026-08-09 10:25:03 +02:00
toxic-agent-demo feat(llm-security): add toxic-agent-demo example for TFA scanner [skip-docs] 2026-05-05 15:15:04 +02:00