ai-agent-supply-chain-attack

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides functional code snippets for Remote Code Execution via insecure deserialization, specifically targeting the Python pickle module and ML framework checkpoints (e.g., PyTorch .pt files). Evidence found in payloads.md and test-cases.md (TC-AISC-001).
  • [COMMAND_EXECUTION]: Example payloads in payloads.md and test-cases.md utilize os.system and shell piping (e.g., curl ... | sh) to execute system commands during model loading or tool invocation.
  • [DATA_EXFILTRATION]: Several scripts within payloads.md demonstrate how to harvest and exfiltrate environment variables (specifically mentioning OPENAI_API_KEY) to external domains like attacker.example using urllib.request and curl.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface. It ingests untrusted data from external sources via WebSearch, WebFetch, and the Hugging Face Hub API, and possesses high-privilege capabilities including Python and Bash execution.
  • Ingestion points: Hugging Face model metadata, web search results, and remote document fetching via huggingface_hub and WebFetch.
  • Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded prompts in fetched data.
  • Capability inventory: The agent is granted Bash, Python, Write, and Edit tools, allowing for system-level actions based on processed data.
  • Sanitization: There is no evidence of sanitization or filtering of external content before processing.
  • [DYNAMIC_EXECUTION]: The skill documents and provides payloads using dynamic execution primitives like eval(), exec(), and pickle.load() to facilitate arbitrary code execution in the context of ML model serving and development (e.g., Keras Lambda layer injection).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 04:36 PM