We do not accept AI-generated security reports. We receive a large number of these and do not have the resources to review them all. Submitting one may result in a ban from the project.
Athena is an AI-powered coding agent that runs locally on your machine. It provides an agent loop with access to powerful tools including shell execution, file operations, and network access.
Athena does not sandbox the agent by default. It runs with the permissions of the user and process that launched it. See packages/coding-agent/docs/containerization.md for supported isolation patterns (Gondolin extension, plain Docker, OpenShell).
If you need true isolation, run Athena inside a container or VM.
Athena's RPC mode is opt-in and intended for local process integration over a framed byte protocol. It is the integrator's responsibility to control which processes can connect to an RPC session; exposing it over a network without authentication is not a supported configuration.
| Category | Rationale |
|---|---|
| Damage from unrestricted tool use | The permission-less default is documented behavior; use containerization for isolation |
| LLM provider data handling | Data sent to your configured LLM provider is governed by their policies |
| MCP server behavior | External MCP servers you configure are outside our trust boundary |
| Malicious config/extension files | Users control their own config and extensions; a user running their own malicious extension is not a vulnerability |
We appreciate your efforts to responsibly disclose your findings, and will make every effort to acknowledge your contributions.
To report a security issue, please use the GitHub Security Advisory "Report a Vulnerability" tab.
The maintainer will send a response indicating next steps in handling your report, and will keep you informed of progress towards a fix and disclosure.