The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection surface identified. The skill ingests untrusted data from external blockchain RPC nodes via methods such as watchContractEvent, getLogs, and readContract (documented in examples/watch-events/README.md and examples/read-contract/README.md).\n
Ingestion points: Blockchain events, logs, and contract return values are retrieved and processed into the agent's context.\n
Boundary markers: There are no instructions or examples showing the use of delimiters or explicit warnings to ignore embedded instructions when handling these external inputs.\n
Capability inventory: The skill facilitates high-privilege operations, including the ability to sign and send transactions (writeContract in examples/write-contract/README.md) and deploy new contracts (deployContract in examples/deploy-contract/README.md).\n
Sanitization: The provided templates and examples do not include logic for sanitizing or validating the raw data ingested from the blockchain before it is presented to the agent.

viem