Launching today
Elva
Goodbye, Postman. Your APIs have new consumers
135 followers
Goodbye, Postman. Your APIs have new consumers
135 followers
Developers use your APIs. Now agents do too. Elva discovers APIs from code, lets you choose what each audience can use, and runs your MCP servers with auth and analytics. Track agent activity and review changes as your code evolves
Theneo
Hey Product Hunt, Ana here 👋
Four years ago, I was fed up with Postman and wanted to replace it. I started with docs as a side project. That became Theneo. We got into YC, and today 20,000+ companies use it.
Through Theneo, we’ve seen how messy API management gets. Now some of our customers get more API calls from agents than humans, while the way they manage those APIs has barely changed.
That’s why we built Elva. We started with a few customers, and today we’re opening it up to everyone.
You can connect your repo and Elva builds your API catalog, maps how it’s used, and checks AI readiness (fyi no spec required).
From there, create contracts around actual use cases: choose the endpoints and fields a customer, partner, or agent needs. Publish to docs platforms like Theneo or Mintlify, mocks, or hosted MCP servers. When your code changes, review the updates or sync them automatically under your publishing rules.
For MCP, we handle hosting and auth. You can test across models, see which tools are slow or failing, and collect feedback directly from agents when they get stuck.
Would love for you to try it with your own repo. Does it cover how you actually manage your APIs? What are we missing?
Solid MCP host. When one server serves different audiences, how do you keep auth and usage analytics separate?
Theneo
@pratiksharda Thanks! We treat each audience as a distinct tenant or client boundary, even when they share the same MCP server. Each gets its own authentication policy, scopes, and credentials, with authorization enforced per tool and resource, not just at the server level.
For analytics evry request carries a tenant/client identifier through the full execution path. That lets us track usage, latency, errors, and quotas separately without exposing sensitive user data. We also keep audit logs scoped by tenant and apply strict redaction and retention policies. In short: shared infrastructure, isolated identity, permissions, quotas, and observability.
Just tested it and it's very smooth. The MCP server generation especially, hadn't seen that anywhere else. Does the tool set scale well for an API with a lot of endpoints, or does an agent get overwhelmed?
Theneo
@maia_mefarishvili Great question, honestly it depends on the model as well, some models e.g GPT Astra, Fable, handle large amount of tools very well and some other models don't. Sometimes we recommend splitting mcp into multiple catalogs, based on use cases. But great thing also is that, we have this feature that allows you to test commands on different models, so you would know exactly if some models are having hard time reading large amount of endpoints
@ana_robakidze Splitting into multiple catalogs by use case makes sense. Is that something you configure manually, or does Elva suggest the split for you?
Theneo
@maia_mefarishvili you can split it via our AI assistant, but for now we do not suggest that automatically, though that is very nice feature, writing it down. Give it a try and let me know if you have any other amazing ideas
As the code base changes, when APIs change, how do the reviews get done: automatically flagging differences, or requiring approval every time?
Btw, Congratulations @ana_robakidze & @mariam_lekveishvili1 ✌️
Theneo
@mariam_lekveishvili1 @aymi_malik great question and answer depends on your configuration. You can either have automatic deployments, e.g code changes -> we automatically scan and receive updates -> if it is connected to an API contract or MCP server without any governance (approval, reviewers etc) then it will be automatically updated.
But if you have any type of governance in place, let's say reviewers, approvals, then they will first receive a notification that API has drifted and ask for your review.
When it comes to partner and public facing APIs we highly recommend adding some kind of governance in place. That way also you can notify right stakeholders (which we have this option as well)
Mailwarm
Do you support agent keys for different agents, so you can see which agent did what?
Theneo
@naimz hey yes we do track which agent and user is using tool and how, which tools are failing, and even if something fails you get a feedback from the agent directly
It is pretty straightforward to set up and get running. I like the MCP setup and the way it lets you test how the API works with agents. Seems like a useful tool for teams working on making their APIs agent-ready. Congrats on the launch!
Theneo
@sophie_gurjishvili Thanks Sophie, let us know how it holds up in production
Theneo
@sophie_gurjishvili thanks so much! I would love to hear more when you get a chance to thoroughly explore it
MCP is very smooth. I also really like the API spec grader, it really gives you an idea about what to improve and in what direction. Good luck on your launch! I will definitely be using this later.
Theneo
@elene_kv thanks Elene, let us know how it holds up in production
Theneo
@elene_kv thanks ! I look forward to hearing your feedback! Please let me know if you have any questions