budget-optimizer
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (
scripts/budget-optimizer.py). This script is used to compute efficiency metrics and generate optimization scenarios based on provided data. - [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection and path traversal due to the ingestion of untrusted user data and local configuration files.
- Ingestion points: User-provided budget and performance metrics; local brand configuration files located at paths derived from user input (
{slug}). - Boundary markers: No delimiters or warnings are used to differentiate untrusted data from system instructions.
- Capability inventory: Execution of local scripts (
scripts/budget-optimizer.py) and reading files from the filesystem (~/.claude-marketing/). - Sanitization: There is no evidence of sanitization for the
{slug}variable, which is used to construct file paths, potentially allowing for path traversal if the agent accesses files outside the intended directory.
Audit Metadata