pm-manipulation
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from external sources, including prediction market comments and social media content (Telegram, X/Twitter, Reddit). \n
- Ingestion points: Market comments are fetched via Polymarket APIs (SKILL.md Step 3), and social media data is collected via WebSearch (SKILL.md Step 4).\n
- Boundary markers: There are no explicit delimiters or instructions to isolate this untrusted data from the agent's core instructions.\n
- Capability inventory: The skill can perform network operations (curl, WebSearch), write to the file system (mktemp, reports), and execute a local notification script (./notify).\n
- Sanitization: No sanitization is performed on the external content before analysis.\n- [COMMAND_EXECUTION]: The skill utilizes shell commands like curl and mktemp to process data and generate reports. This poses a potential risk of command injection if identifiers obtained from external APIs (such as event slugs or tokens) are not properly validated before being used in shell commands.\n- [EXTERNAL_DOWNLOADS]: The skill fetches market data and user comments from Polymarket's public APIs and retrieves news coverage from various international outlets. These operations are essential for the skill's purpose of monitoring market manipulation.
Audit Metadata