profiler-load-data
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads raw text from a file system path and returns it to the agent, creating a surface for indirect instructions. 1. Ingestion points: File contents are read into the agent's context via the
filePathparameter in SKILL.md. 2. Boundary markers: The documentation does not specify the use of delimiters or instructions to the agent to treat the loaded data as untrusted. 3. Capability inventory: The skill allows reading any file on the system readable by the host process, up to 10MB (SKILL.md). 4. Sanitization: None; file contents are returned as raw text without validation. - [DATA_EXPOSURE_AND_EXFILTRATION]: The skill facilitates reading arbitrary files from the local filesystem. The
filePathinput allows the agent to access any file, which could include sensitive configuration or credentials if the agent is misdirected, as no directory whitelisting is documented. - [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The troubleshooting section suggests installing
unity-mcp-clivia npm (npm install -g unity-mcp-cli). This utility is a vendor-owned resource associated with the author 'ivanmurzak' and is necessary for the skill's operation.
Audit Metadata