multi-asset-morning-briefing
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill retrieves market data from official and well-known financial institutions, including the U.S. Department of the Treasury, the Federal Reserve Bank of St. Louis (FRED), Cboe Global Markets, the European Central Bank (ECB), and the Bank of Korea (ECOS).
- [COMMAND_EXECUTION]: The skill instructions involve executing a local helper script (
scripts/market_data.py) and using a vendor-specific CLI tool (@nomadamas/k-skill) for managing skill instructions and updates. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external web sources and APIs. While this creates a surface for indirect prompt injection, the risk is mitigated by the use of structured parsing in the Python script and the lack of high-privilege capabilities.
- Ingestion points: Data is fetched from Treasury, FRED, Cboe, ECB, and ECOS APIs within
market_data.py. News content is gathered via browser tools as described ininstruction.md. - Boundary markers: The skill uses specific parsing logic in
market_data.pyto extract numeric and date values, and thereferences/format-rules.mdfile mandates precise verification, which limits the influence of free-text injection. - Capability inventory: The skill uses a provided Python script (
market_data.py) for data retrieval; no arbitrary shell execution,eval()usage, or sensitive file-write capabilities were found across the skill files. - Sanitization: The Python script performs explicit type conversion (e.g., to float or date) and validation on the retrieved data, ensuring the agent only receives structured financial information.
Audit Metadata