search-first

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill defines a 'Complete Mode' workflow that triggers a subagent using a template. This template interpolates untrusted user input (e.g., function descriptions, language constraints) directly into the instructions for a general-purpose subagent, creating a surface for indirect prompt injection.
  • Ingestion points: User-provided placeholders [功能描述], [LANG], and [ANY] within the subagent task definition in SKILL.md.
  • Boundary markers: The prompt template lacks delimiters (such as XML tags or triple quotes) or explicit instructions for the agent to ignore potentially malicious embedded commands within the interpolated data.
  • Capability inventory: While the search-first skill contains no executable code itself, it delegates tasks to a 'general-purpose' subagent which may have access to tools or capabilities that could be abused through injection.
  • Sanitization: There is no evidence of sanitization, escaping, or validation of the user-provided placeholders before they are interpolated into the prompt.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 12:38 AM