trading-skills-navigator
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes a bundled Python utility (
recommend.py) to analyze user goals and repository metadata. This script operates locally and is a core component of the skill's functionality. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by processing natural language user goals to generate recommendations.
- Ingestion points: The user's natural language goal is extracted and passed verbatim to the
--queryparameter of therecommend.pyscript as described inSKILL.mdStep 2. - Boundary markers: The command template in the instructions does not use specific delimiters (like triple quotes or XML tags) to wrap the user-supplied query.
- Capability inventory: The recommendation script is highly constrained. It only performs text normalization and keyword matching against static metadata files (YAML and JSON). It lacks the capability to write files, perform network operations, or execute further system commands.
- Sanitization: The script normalizes input by converting to lowercase and collapsing whitespace to facilitate matching against its internal persona table.
Audit Metadata