Lack of input validation for prompt

[ka1hatsu]

Lack of Input Validation for prompt

https://github.com/sendaifun/solana-agent-kit/blob/main/src/tools/agent/get_info.ts

Issue: The prompt parameter is directly used in the messages array without any validation. This could lead to potential misuse or injection attacks if the input comes from an untrusted source.

Example Vulnerability: If a user injects malicious instructions into the prompt, the model could return unintended outputs (e.g. instructions for harmful actions).

Mitigation: Sanitize and validate the prompt input.

For example:

if (typeof prompt !== "string" || prompt.length > 1000) {
  throw new Error("Invalid prompt input");

}

[michaelessiet]

Hey @ka1hatsu , how do you recommend we implement sanitization?