Ask HN-style: How teams handle permissions for AI agents in production
Expensive_Hamster_11 · reddit · 2026-10-20
The poster asks devs and founders building AI agents: when an agent connects to internal tools, APIs, databases, or MCP servers, how do you ensure it only accesses what it should?
Three specifics: are permissions managed per-tool or centrally; how do you audit what an agent accessed and why; and do you test whether an agent can reach data it shouldn't, or is that still manual? They want real production setups and pain points, not theory.
More from coding & agent
- Personal software factories are a waste of time, argues developer — rseroter · 2026-10-20
- Keyfleet lets personal AI agents join permanent fleets via 10–100 USDC onchain Keys — seanwbren · 2026-10-20
- Claude SEO, the 15k-star toolkit, now runs natively in Hermes with 45 SEO skills — MicahBerkley · 2026-10-20
- Testing EverOS: AI Agents Need Experience, Not Just Memory — Aiden_Tech_Ai · 2026-10-20
- Investor shares his internal agent stack: Mastra, Trigger.dev, Context and more — ryanbed · 2026-10-20
- 'Suddenly everyone is building Sandboxes' — agent sandboxes become the new gold rush — mgill25 · 2026-10-20