HackerTrans
TopNewTrendsCommentsPastAskShowJobs

marinerk9

no profile record

Submissions

Show HN: Hydra – Real-time ops dashboard for developers running AI agents

github.com
1 points·by marinerk9·hace 4 meses·2 comments

comments

marinerk9
·hace 4 meses·discuss
This is a sharp distinction and you're right to separate them.

Hydra's scope right now is purely Layer 1 observability. You can see that an agent opened a network connection or spawned a subprocess, but you're watching it happen, not blocking it.

The enforcement layer is where it gets interesting and hard. Process-level sandboxing, network allowlists per agent, file access boundaries that's a meaningfully different architecture than just reading nettop output and rendering it. It has to sit below the agent, not alongside it.

It's on the roadmap but I'd rather ship good observability first and build enforcement on top of real usage patterns, rather than guessing at policy rules upfront. The data Hydra collects (which processes talk to which endpoints, what files get touched, resource consumption patterns) is actually the foundation you'd need to define sensible authorization boundaries.

Appreciate the comment this is exactly the kind of thinking that sharpens the project.
marinerk9
·hace 4 meses·discuss
Update: just pushed a security audit to the repo.

0 known vulnerabilities, Electron sandbox enabled, contextIsolation enforced, openExternal URL allowlisted. Full report in SECURITY.md if anyone wants to dig in.