picks-tracker
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from
memory/logs/, which are files generated by other potentially untrusted skills. This creates an attack surface where malicious content in the logs could influence the agent's output or notification content. - Ingestion points: Reads files from the
memory/logs/directory. - Boundary markers: None explicitly used to separate untrusted data from the agent's instructions, though it uses specific text patterns (
**Token:**,**Market:**) to extract data. - Capability inventory: The skill possesses network access (
curl), file writing capabilities (output/articles/), and notification execution (./notify). - Sanitization: The skill relies on simple extraction logic; it does not perform advanced validation or sanitization of the values extracted from logs before they are used in notifications or file writes.
- [COMMAND_EXECUTION]: The skill uses
curlto interact with external APIs for price and market data retrieval. - Evidence: Executes shell commands to query the CoinGecko API (
api.coingecko.com) and Polymarket API (gamma-api.polymarket.com).
Audit Metadata