SuperAGI · Issues· 263 open

Security: preferred private channel to report a vulnerability?

[SECURITY] Remote Code Execution via Celery pickle deserialization over an unauthenticated Redis broker

I ran the following, but there was no code in the document when apparently it was done.

Feature Request: Add Moorcheh as vector store backend for agent LTM

[Bug] Path Traversal leading to Arbitrary File Write/RCE in ReadEmail Tool

Task/step output handlers run eval() on raw model output (code execution via model reply)

[SECURITY] Critical RCE via eval() on LLM output

Long term memory

Security: eval() on marketplace data and LLM output enables code injection

Runtime Open Federation — federate your SuperAGI agents, earn USDC on Base chain

Federation Partnership — Runtime Open Federation (open task market + USDC)

SuperAGI execution configuration cross-organization disclosure

Cross-Organization Agent List Disclosure via Project ID

Cross-Organization Agent Scheduling Without Authorization Check

Cross-Organization Agent Lifecycle Actions Without Authorization Check