use-tinyfish-cli
Warn
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The instruction "Match the user's lang˝uage" contains an obfuscated character (U+02DD, Double Acute Accent) within the word "language". This technique is commonly used to evade static analysis or keyword-based security filters.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to fetch and process content from arbitrary external URLs. This untrusted data could contain malicious instructions that the agent might inadvertently follow.
- Ingestion points: Web content retrieved via the
tinyfish fetch content getcommand. - Boundary markers: The instructions do not define delimiters or warnings to separate external data from system instructions.
- Capability inventory: The skill has access to shell command execution via the
tinyfishCLI tool. - Sanitization: There is no evidence of sanitization, filtering, or escaping procedures applied to the fetched content before it enters the agent context.
- [COMMAND_EXECUTION]: The skill utilizes the
tinyfishCLI tool, allowing the agent to execute shell commands to perform web searches and data extraction.
Audit Metadata