lov-expense-report
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted invoice images and text descriptions provided by the user. While the primary purpose is data extraction, these inputs could contain malicious instructions designed to influence the agent's behavior during the categorization or confirmation steps.
- Ingestion points: Step 1 in
SKILL.md(Read tool for images, parsing text descriptions). - Boundary markers: Absent. The skill does not explicitly instruct the agent to ignore instructions embedded within the invoice content.
- Capability inventory: Executes local Python script (
generate_report.py) and writes to the file system. - Sanitization: The skill relies on the agent's internal extraction logic; no explicit sanitization of text from images or user input is described before it is structured into JSON.
- [COMMAND_EXECUTION]: The skill executes a local Python script
generate_report.pyas part of its normal workflow to generate Excel reports. This is a core functionality and uses a localized path. - Evidence:
python expense-report-skill/scripts/generate_report.py --input /tmp/invoices.json --output ...inSKILL.md.
Audit Metadata