Skip to content
Coming soon

Expertise

When I review an API specification, an agent skill or a process description, I check it against standards I've written down for myself. This service turns those checks into something your agent or CI pipeline can call: you send an artifact and get back the findings I'd give an employer or client.

The answers come from rules I wrote and maintain, not from a language model, so the same input gets the same result. Nothing is live yet: this page describes what I'm building, and details may still change.

Three checks

  • API governance check

    €10 per artifact
    You give
    An OpenAPI 3.0 or 3.1 document.
    You get
    Findings with a severity, the JSONPath where they apply, the issue and a fix, plus a governance score per dimension.
    Up to 50 operations. Each further 50 operations adds €5.

  • Skill design check

    €5 per artifact
    You give
    An Agent Skill folder: SKILL.md and the files next to it.
    You get
    Findings on structure, triggering, scope, progressive disclosure and safety, by file and line, plus a score.

  • Process description review

    €15 per artifact
    You give
    A process description in the published process format.
    You get
    Findings on roles, handoffs, controls, exceptions and measurability, plus a score.

Planned prices in euros, excluding VAT.

Three ways in

  • MCP, for agents

    Sign in with OAuth and the server offers the checks as tools, with skills that tell your agent when to use them.

    https://mcp.brandonverzuu.com/mcp

  • The CLI, for people and coding agents

    bv-expertise, from the npm package @brandonverzuu/expertise. It validates your file locally first, so a malformed one never costs anything, and writes text, JSON, SARIF or Markdown.

    bv-expertise check api openapi.yaml

  • REST, for CI

    The same checks over HTTP with an account API token. In a pipeline the CLI does this for you, can fail the build on errors, and its SARIF output shows findings in GitHub code scanning.

    POST /v1/checks/{kind}

Skills and the process format

Public skills tell your agent when a check is worth running, how to prepare the input and how to read the result. They don't contain the rules. They'll be published in this site's public repository (opens in new tab) and served over MCP.

The process review needs your process in a structured format. I'm publishing it as a JSON Schema at /schemas/process-description/v1.json. The skill helps your agent turn a prose description into it, and asks you for anything that's missing instead of making it up.

Paying per check

You pay for each check as you run it. There are no packs and no balance. You see the price for your file first, and a check that fails on my side isn't charged.

Payment goes with the request itself: over x402, in EURC or USDC from a wallet linked to your account, or over Stripe's Machine Payments Protocol in euros. Your agent or the CLI asks you before it pays. There's no free trial.

Pricing →Terms →Privacy →