crsp-lseg-splice
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/crsp_lseg_splice.pyperforms SQL queries against the WRDS (Wharton Research Data Services) PostgreSQL database and executes data retrieval calls through the official LSEG Refinitiv Python library. These operations are essential to the skill's core purpose and are implemented using secure coding patterns. - [EXTERNAL_DOWNLOADS]: The skill retrieves financial market data from well-known and trusted academic and commercial platforms, including WRDS and LSEG Refinitiv. These network operations are documented and consistent with the skill's stated intent.
- [INDIRECT_PROMPT_INJECTION]: The skill processes financial data from external database and API sources, which serves as a theoretical ingestion surface for indirect prompt injection if the sources were compromised. 1. Ingestion points: Data is retrieved into Python DataFrames via
psycopg2andlseg.datamethods inscripts/crsp_lseg_splice.py. 2. Boundary markers: The skill handles structured, numeric financial data rather than natural language, which naturally restricts the injection surface. 3. Capability inventory: The skill is capable of network communication (data retrieval) and local filesystem writes (saving Parquet files). 4. Sanitization: The skill employs parameterized SQL queries to prevent injection attacks at the database interface layer.
Audit Metadata