How do you separate agent identity from tool authorization in MCP setups?
PuzzleheadedNet2204 · reddit · 2026-09-22
The author observes a shared pain point across agent stacks: the model/runtime can call tools, but auth is still basically a long-lived shared API key. They ask the community how people split three things in practice:
- Workload/agent identity: how do you identify the specific agent instance making the call (short-lived credentials, SPIFFE-ish workload identity, something else)?
- Per-action authorization: once identity is established, do you re-check scopes on every tools/call, or is that baked into the session?
- Discovery vs invocation: do you allow anonymous tools/list, or require auth before listing capabilities?
Especially interested in multi-agent, multi-MCP-server setups—and what broke first when moving past a single shared key.
More from coding & agent
- mcp-server-s3 ships MCP server letting agents browse, upload and share S3 files — modelcontextprotocol · 2026-09-22
- apibase unifies 327 tools from 92 providers behind one pay-per-call MCP endpoint — modelcontextprotocol · 2026-09-22
- Dev built an AI game-playing plugin but shelved it: vision models too slow and costly — ezshine · 2026-09-22
- Developers debate persistent file storage options for AI agents across runs — OwlZealousideal4779 · 2026-09-22
- Laya: 11.6k-star open-source engine outputs typed decisions in 33ms, no generation — pandeyparul · 2026-09-22
- Dev open-sources Jev-powered search CLI to cut token burn from LLM web research — gaganghotra_ · 2026-09-22