blockchain-sql
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external CSV files, creating a potential vector for indirect prompt injection.
- Ingestion points: The script
scripts/blockchain_sql_diagnostics.pyreads user-supplied CSV files using thepandas.read_csvfunction. - Boundary markers: No specific delimiters or "ignore embedded instruction" prompts are provided to the agent to treat the script output as potentially untrusted data.
- Capability inventory: The skill is configured to execute local Python scripts and write JSON artifacts to the local filesystem.
- Sanitization: The diagnostic script provides mitigation by coercing values to numeric or datetime types; however, CSV column headers are extracted verbatim into the summary JSON, which could be used to pass instructions to the agent if headers are attacker-controlled.
Audit Metadata