Add Aevrin MCP to your agent once, then just ask. It checks an MCP server before your agent trusts it, inspects a skill before you install it, shows what your agent can reach, finds and fixes problems in your code, reviews pull requests, and writes tests that really run.
Free plan, no card. Every product is included.
Each of those needs a safety check.
Aevrin MCPOne install. Six safety checks.
Aevrin MCP is a plug for Claude Code, Codex, Cursor or any app that speaks MCP. Run these two commands, start a new session, and ask your agent for a check in plain words.
aevrin login
claude mcp add-json --scope user aevrin '{
"type": "http",
"url": "https://api.mcp.aevrin.net/mcp",
"headersHelper": "aevrin mcp-header"
}'An MCP server is a plug that gives an AI agent new tools. Aevrin runs the server in a safe sandbox, reads every tool it offers, and grades it from A to F.
A and B mean allow. C and D mean ask a person first. F means block. Your agent can follow that rule directly.
If a server hides its tools, the scan says it could not check them. An unfinished scan is marked incomplete and gets no grade. It is never called clean.
Each problem is matched to the OWASP MCP Top 10, a public list of the biggest MCP risks. It reads the same in the dashboard, the terminal and CI.
A skill is a set of instructions and code your agent will follow. Point Aevrin at one and it reads both. Then it tells you: Safe, Caution or Do not install.
A trick can hide in plain sentences, not just in code. Aevrin reads what the skill tells your agent to do, next to what its scripts really do.
Code checks, security rules and a lookup of known library bugs run first. An optional AI pass then reads for intent that rules cannot see.
SARIF for code scanning, JSON for scripts, Markdown for a pull request, and a short summary in your terminal.
Aevrin reads how an AI coding agent is set up: its MCP servers, skills, plugins and hooks, and the files, keys, commands and network it can reach. Then it gives the setup a score out of 100.
Every agent starts at 100. Each point it loses has a name, such as a key the shell can read, so you know which setting cost it.
Aevrin joins what an agent can run with what it can read. You see the route from the agent to a secret, not a flat list of permissions.
An MCP server your agent can call keeps the grade from its own scan. A server nobody has scanned is shown as ungraded, never as safe.
Connect Claude Code, Codex, Cursor or another MCP app to Aevrin MCP. Your agent can then scan code, explain a problem, fix it and scan again, all in the same chat.
Aevrin MCP carries checks for web apps, APIs, logins, injection, cloud, containers, AI agents, smart contracts and more. Your agent finds the right one by asking.
These checks run on your machine with your own permissions. Aevrin gives your agent the method. Your code is not uploaded.
Each problem says where it is, what sets it off and what it lets someone do. It is checked against the code before you see it.
Aevrin reviews pull requests, code changes and whole repositories for problems an attacker could use. It shows how the attack would work and suggests a safe fix you can apply and scan again.
Some problems depend on what the code means, not how it looks, like a missing check on who owns a record. The reviewer reasons about that.
Each problem names the file and line, the input that reaches it, and what an attacker gets. A person can confirm it in minutes.
The fix is applied as a change and the code is scanned again. The pull request closes with the problem gone, not just commented on.
Aevrin finds the code a change left untested and writes tests in your project's own style. It runs them, fixes the ones that fail, and gives back only tests that pass and add coverage.
Every test runs before you get it. A failing test is fixed or dropped. It is never handed back as if it worked.
Tests use the framework, layout and style your project already has, so they read like your team wrote them.
You see which lines the new tests reach and how much coverage went up, so you can tell useful tests from filler.
Every scan grades a server from A to F, matches its problems to the OWASP MCP Top 10, and says which step stopped if it could not finish.
Each server gets a letter from A to F and a risk score out of 100. Higher is worse.
A scan has five steps. If one cannot finish, the report names it and gives no grade.
Problems map to 7 of the 10 groups. MCP08 is not tested, and every report says so.
--fail-on picks the level that fails a build. Exit 3 means incomplete, so a half-done scan cannot pass quietly.
Pro can export a printable report. What was not checked stays in it, so a saved copy cannot make a partial scan look clean.
An AI explanation says what a problem means. It never changes a grade.
The free plan has every Aevrin product and 1,500 credits a month. No card, and nothing renews on its own.
Aevrin reads, scans or plugs into each of these. You keep working where you work today.
If an app can use MCP servers, it can use Aevrin MCP.
Your agent uses the same Aevrin key as the CLI. There is no second account.
Your agent asks for what it needs. Aevrin answers with checks an Aevrin admin has reviewed.
Ask your agent, type a command, add a step to CI, or open the dashboard. A problem reads the same in every place.
Aevrin MCP gives Claude Code, Codex or Cursor three tools: find a check, get it, run it.
The aevrin command scans a server, a folder or your agent setup.
One step in your pipeline. It fails the build at the level you choose.
Every scan, problem and agent in one place. Your account is yours alone.
Two plans: Free and Pro, both spending one monthly pool of credits. Pro has a far bigger pool and adds AI code review and report export. It buys one month or one year, and nothing renews on its own.
$0a month
No card needed. Nothing to cancel.
For trying Aevrin on the servers and code you use.
Start freeIncludes
$17a month
$199 paid once for 12 months. You save $29 a year.
For daily use, in your terminal and in CI.
Choose Pro, annual billingIncludes
Something not answered here? Write to support.
Scanning
Agent posture
Aevrin MCP
Plans and billing
