comparable-analysis-tp

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill uses a path traversal pattern to access a file outside its immediate directory: '../../CLAUDE.md'. While this is common for project-level instructions in certain environments, it allows the agent to access data beyond the scope of the skill folder.
  • [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface due to its dependency on external instructions from a configuration file. 1. Ingestion points: The file '../../CLAUDE.md' (referenced in the '加载上下文' section) is used to define industry, geographic, and financial benchmarks. 2. Boundary markers: Absent. The skill does not define specific delimiters or instructions to prevent the agent from obeying malicious commands potentially embedded in the config file. 3. Capability inventory: The skill utilizes the agent for calculating financial ratios and generating formatted reports; no high-privilege tool use (such as shell execution, file modification, or network operations) is defined within the skill file. 4. Sanitization: Absent. There are no instructions to validate, filter, or escape the content retrieved from the external configuration file before it is used by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 02:26 AM
Security Audit — agent-trust-hub — comparable-analysis-tp