The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to recognize and act upon error signatures received from external blockchain services. Because these error strings are processed to trigger specific agent actions, they represent a surface for indirect prompt injection.
Ingestion points: Error signatures and messages from Polymarket, Relay, Hyperliquid, and Solana RPC calls in SKILL.md.
Boundary markers: The instructions do not define clear delimiters or "ignore instructions" warnings for the error content being analyzed.
Capability inventory: The skill directs the agent to perform GET requests (for wallet and account state) and state-changing POST requests (specifically to /agent/trading/abstraction) based on the content of these errors.
Sanitization: There is no evidence of data sanitization or integrity checks for the error messages before the agent executes the mapped fix actions.

openfin-troubleshooting