finta-local-dev-loop
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external data from local CSV files.
- Ingestion points: The Python analysis blocks in
SKILL.mdusepd.read_csv("pipeline-export.csv")to load external data into the agent's memory. - Boundary markers: No specific boundary delimiters or 'ignore' instructions are provided for the CSV content.
- Capability inventory: The skill utilizes
Bash(python3:*)to perform data analysis and report generation. - Sanitization: While programmatic sanitization is not in the code, the skill instructions explicitly mandate the use of de-identified data, isolated project environments, and the exclusion of PII (names, emails) from generated reports.
Audit Metadata