LLM oracle generator can give empty response to task request and LLM oracle validator can validate it causing unfair rewarding system and losses of funds for users.
In the respondfunction in LLMOracleCoordinator.solthere is no validation check of the response given by the LLM oracle generator. With this, malicious LLM oracle generator can find task with 0 parameters.numValidationsand give spam response and get allowance increase. Even the task with parameters.numValidationsgreater than 0, there could be malicious LLM validator oracle that validates this empty response working in cohort with the malicious LLM oracle generator.
Malicious LLM oracle generator could spam empty response and another malicious LLM oracle validator can validate the empty response which causes unfair rewarding system and losses of funds for users.
Below is a snippet of the respondfunction:
Here, the outputvariable is the response by the LLM oracle generator. As seen above, the outputis never checked/validated in this function.
Manual review
Check the length of outputin the respondfunction so that empty output cannot be accepted.
The contest is live. Earn rewards by submitting a finding.
This is your time to appeal against judgements on your submissions.
Appeals are being carefully reviewed by our judges.