{
 "public_id": "AIR-2026-0103",
 "incident_id": "2026-05-gecko-skill-scanner-test-file-bypass",
 "date_disclosed": "2026-05-07",
 "date_occurred": null,
 "title": "Agent skill scanners bypassed by hiding malicious code in bundled test files",
 "class": "research_demo",
 "attack_surface": "skill_plugin",
 "initial_vector": "malicious_payload_outside_scanner_threat_model",
 "agent_capability_abused": [
  "test_runner_execution",
  "local_permissions"
 ],
 "target_product": "Snyk Agent Scan, Cisco AI Agent Security Scanner, VirusTotal Code Insight",
 "target_vendor": "Snyk / Cisco / Google (VirusTotal)",
 "autonomy_level": "semi_autonomous",
 "impact_realized": false,
 "impact_type": [
  "arbitrary_code_execution",
  "credential_theft"
 ],
 "blast_radius": "all agent-skill scanning that inspects only SKILL.md and agent-invoked scripts",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI05",
   "ASI04"
  ],
  "mitre_atlas": [],
  "labels": [
   "scanner_evasion",
   "skill_marketplace",
   "post_approval_modification"
  ]
 },
 "defense_present": "three commercial/automated agent-skill scanners",
 "defense_bypassed": true,
 "fix_shipped": "disclosed to vendors; scanners' scope described as incomplete rather than broken",
 "primary_sources": [
  "https://venturebeat.com/security/anthropic-skill-scanners-passed-every-check-malicious-code-test-file"
 ],
 "verified_quote": "The scanner is not broken. It is incomplete.",
 "confidence": "high",
 "notes": "Research by Jeevan Jutla of Gecko Security. Malicious code in *.test.ts / conftest.py runs under the normal test runner with full local permissions; the scanners' detection categories target the agent-interaction layer, not the developer toolchain. Also enables a push-clean-then-add-test-file post-approval attack. Article cites a SkillScan academic audit finding 26.1% of 31,132 skills contained vulnerabilities.",
 "risk_category": "security_fraud",
 "risk_secondary": [],
 "risk_subcategory": "Supply Chain Compromise",
 "risk_sub_secondary": [
  "Credential & Secret Theft",
  "Remote Code Execution & Sandbox Escape"
 ],
 "ai_entity": "generative",
 "agency": "agent_acted",
 "agency_debatable": true,
 "agency_reason": "the bundled test code runs with the agent's local permissions",
 "vector_family": "malicious_component",
 "vector_detail": "malicious_payload_outside_scanner_threat_model",
 "guardrail_kind": [
  "unknown"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [],
  "atlas": [],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI05",
 "agent_framework": "unknown",
 "model_provider": "google",
 "model_name": "",
 "detection_mechanism": "external_researcher",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "supply_chain_vulnerabilities",
 "eai_domain": "tool_misuse",
 "eai_secondary": [
  "credential_theft",
  "privilege_escalation"
 ],
 "eai_reason": "subcategory: Supply Chain Compromise",
 "asi_curated": [
  "ASI05"
 ],
 "asi_scope": "in_scope",
 "asi_provenance": "curated_plus_rule",
 "asi_reason": "curated: ASI05 | added ASI04 (vector_family=malicious_component)",
 "submitted_by": "EAI Internal"
}