geo-human-editor
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions and documentation encourage the execution of a local Python script
scripts/humanize_content.pyto process text files. This script performs simple string replacements to "humanize" content and operates locally without external network dependencies. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and audit external content provided by users, which represents a potential surface for indirect prompt injection.
- Ingestion points: Content provided in the
--inputfile for the humanization script or passed directly to the agent via the prompt is evaluated for tone and quality. - Boundary markers: The instructions do not define specific delimiters or "ignore" instructions to separate the untrusted input text from the agent's operational logic.
- Capability inventory: The provided Python script
scripts/humanize_content.pyutilizes the file system to read input and write output based on arguments. The agent also performs text analysis and rewriting as part of its core function. - Sanitization: The script performs basic string replacements for vocabulary and formatting (e.g., stripping markdown bold markers) but does not implement validation or escaping to prevent the processing of instructions that might be embedded in the audited text.
Audit Metadata