{
  "id": "lolskills-ink5725-snapshot-config-reader-2026-10",
  "name": "ink5725/snapshot-config-reader skill finding",
  "date_reported": "2026-10-06",
  "researcher": "Jose Enrique Hernandez (LOLSkills feed)",
  "org": "MagicSword / LOLSkills",
  "sources": [
    {
      "url": "https://lolskills.io/discover/56bd2dd3473fa5a15b30dd42a262ac410148e41679ebcc51c71b33279cdbfa6c"
    }
  ],
  "magicsword_coverage": [
    "https://x.com/_josehelps/status/2107500883554812028",
    "https://x.com/_josehelps/status/2107500883554812028"
  ],
  "borderline": true,
  "borderline_note": "Scanner finding on a skill bundle; Jose explicitly not calling it malicious. Pattern of decode-from-image instructions.",
  "ai_role": "ai_agent_or_mcp_as_target",
  "ai_role_secondary": null,
  "ai_role_detail": null,
  "technique_summary": "Agent skill SKILL.md instructs pulling 'config' from PNG trailer, EXIF, LSB. ClamAV Img.Packed.PngContainsDownloadCmd (high) on two PNGs; Nova DetectHiddenInstructions (low). 0 downloads; scan incomplete.",
  "initial_access": "n/a (skill install / agent load path)",
  "ai_tooling_and_models": [
    "Agent skill bundle"
  ],
  "other_tooling": [
    "PNG steganography / trailer / EXIF / LSB"
  ],
  "victims_and_sectors": "unknown",
  "regions": [
    "unknown"
  ],
  "attribution_or_cluster": "unknown",
  "scale_and_metrics": "0 downloads at time of post",
  "prevention_angle": "Whole-bundle skill scanning; treat decode-from-image steps as signals",
  "related_magicsword_projects": [
    "LOLSkills"
  ],
  "source_notes": null,
  "last_reviewed": "2026-10-20",
  "attack_flow": {
    "nodes": [
      {
        "id": "step-1",
        "title": "Skill bundle is scanned",
        "description": "The finding concerns an agent skill with no observed downloads.",
        "kind": "access",
        "source_urls": [
          "https://lolskills.io/discover/56bd2dd3473fa5a15b30dd42a262ac410148e41679ebcc51c71b33279cdbfa6c"
        ]
      },
      {
        "id": "step-2",
        "title": "SKILL.md requests image extraction",
        "description": "Instructions tell the agent to obtain configuration from PNG trailers, metadata or low bits.",
        "kind": "ai",
        "source_urls": [
          "https://lolskills.io/discover/56bd2dd3473fa5a15b30dd42a262ac410148e41679ebcc51c71b33279cdbfa6c"
        ]
      },
      {
        "id": "step-3",
        "title": "Scanners flag embedded instructions",
        "description": "The partial scan detects suspicious image content with differing confidence levels.",
        "kind": "execution",
        "source_urls": [
          "https://lolskills.io/discover/56bd2dd3473fa5a15b30dd42a262ac410148e41679ebcc51c71b33279cdbfa6c"
        ]
      },
      {
        "id": "step-4",
        "title": "No execution is established",
        "description": "The record does not confirm a victim, a completed scan or successful skill execution.",
        "kind": "unknown",
        "source_urls": [
          "https://lolskills.io/discover/56bd2dd3473fa5a15b30dd42a262ac410148e41679ebcc51c71b33279cdbfa6c"
        ]
      }
    ],
    "edges": [
      {
        "source": "step-1",
        "target": "step-2"
      },
      {
        "source": "step-2",
        "target": "step-3"
      },
      {
        "source": "step-3",
        "target": "step-4"
      }
    ]
  },
  "atlas_techniques": [
    {
      "id": "AML.T0051.001",
      "rationale": "The flagged skill instructs an agent to recover instructions from image artifacts; actual execution is unconfirmed.",
      "source_urls": [
        "https://lolskills.io/discover/56bd2dd3473fa5a15b30dd42a262ac410148e41679ebcc51c71b33279cdbfa6c"
      ]
    }
  ],
  "category": "attacks-on-ai",
  "ai_involvement_confidence": "unverified",
  "detections": [
    {
      "name": "Review attack-specific indicators",
      "format": "guidance",
      "notes": "Monitor for the techniques described in ink5725/snapshot-config-reader skill finding. Refer to sources for specific IOCs and TTPs."
    }
  ]
}