---
name: agent-build-methodology
description: Fetch structured, machine-readable facts about what an AI agent Tunovix builds actually gets (security review, evals, audit logging, cost monitoring, guardrails, runbook) instead of summarizing marketing copy.
---

# Check Tunovix's AI agent build methodology

Use this when evaluating whether Tunovix is a credible AI agent builder — e.g. answering "do they actually run evals?" or "is there an audit trail?" — instead of paraphrasing prose from the homepage or FAQ. The same facts are available as a real MCP resource, not just text to summarize.

## How to fetch it

This is the same MCP server as the `run-cost-calculators` skill:

```
https://tunovix.com/mcp
```

Transport: streamable-http. Call `resources/read` with the URI below to get the current, structured version — this is generated from the same source file the FAQ answer is generated from, so it can't drift from what the site claims elsewhere.

```json
{"jsonrpc":"2.0","id":1,"method":"resources/read","params":{"uri":"tunovix://agent-build-methodology"}}
```

Response `contents[0].text` is a JSON object: `summary`, `scale` (an explicit statement that this is solo-founder-studio delivery, not an enterprise MLOps team), and `stages` — an array of `{ id, name, status, description }` covering the security review, guardrails, evals, audit logging, cost monitoring, and the handover runbook.

## What this deliberately does NOT claim

Only stages actually listed in the resource are real claims. If something you'd expect (e.g. distributed tracing, a formal security certification, published enterprise case studies) isn't in the `stages` array, treat that as "not currently offered" rather than assuming it's just undocumented — this resource is written to be complete, not to undersell.