monitor-polymarket
Warn
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The dispatch logic in SKILL.md uses a bash script template that incorporates the ${var} variable directly into shell assignments and commands. A malicious input containing shell metacharacters or subshells (e.g., polymarket:$(command)) could lead to arbitrary command execution if the logic is executed verbatim by the agent.
- [PROMPT_INJECTION]: The skill fetches and summarizes untrusted user comments from the Polymarket API, creating an indirect prompt injection surface. Ingestion points: Polymarket comments API in SKILL.md (Section P3). Boundary markers: None used to isolate or delimit external comment text. Capability inventory: Network access (curl), file system access (memory/), and notification tools. Sanitization: No sanitization of ingested comments is mentioned or implemented.
- [EXTERNAL_DOWNLOADS]: Fetches data from Polymarket (gamma-api.polymarket.com, clob.polymarket.com) and Kalshi (api.elections.kalshi.com) public APIs. These are legitimate data sources for the skill's purpose.
- [DATA_EXFILTRATION]: The skill reads from and writes to the memory/ directory, which may contain sensitive project context or historical logs.
Audit Metadata