14 days ago
I'm concerned about the security of my production environment on Railway when working with it through AI agents. The main concern is that an AI agent could delete production data or disrupt a service. Therefore, I'd like to configure an API token with read-only permissions for it. As far as I understand, Railway doesn't provide this capability?
Is there a recommended way to work with Railway through AI agents? How is it recommended to grant them access to the Railway production environment?
Pinned Solution
14 days ago
Unfortunately, no, Railway does not have tokens with read-only limits.
Railway also has an MCP server you can use: https://docs.railway.com/ai/mcp-server
I'd also recommend creating custom instructions that require the agent to ask for your permission first before changing production configurations or performing destructive actions.
1 Replies
14 days ago
This thread has been opened as a bounty so the community can help solve it.
Status changed to Open Railway • 14 days ago
14 days ago
Unfortunately, no, Railway does not have tokens with read-only limits.
Railway also has an MCP server you can use: https://docs.railway.com/ai/mcp-server
I'd also recommend creating custom instructions that require the agent to ask for your permission first before changing production configurations or performing destructive actions.
Status changed to Solved 0x5b62656e5d • 14 days ago