clade-model-inference

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to use the @claude-ai/sdk package for Node.js implementations. This is not the official Anthropic SDK (which is @anthropic-ai/sdk). Directing users or agents to install non-official packages increases the risk of supply chain attacks or using unvetted code.\n- [PROMPT_INJECTION]: The skill implements a pattern in Step 3 of SKILL.md where model output is directly parsed as JSON and used in logic. This represents a surface for indirect prompt injection if the model processes untrusted data.\n
  • Ingestion points: LLM output retrieved via message.content[0].text in SKILL.md.\n
  • Boundary markers: None; the system prompt does not include markers to distinguish instructions from data.\n
  • Capability inventory: The skill uses JSON.parse() to process the output, which then controls subsequent logic flow.\n
  • Sanitization: There is no validation or sanitization of the model's output before it is parsed or used.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 01:00 PM
Security Audit — agent-trust-hub — clade-model-inference