Connect your AI assistant

Ask the AI assistant you already use to find a standard, name the misconceptions that get in its way and write How the AI helps with you. It takes one address, with nothing to install and no key to get.

https://api.viablelab.org/mcp

This is an MCP server. To have your AI build courses and agents in VIABLE as you, use the CLI instead.

Add the address

Claude Code:

claude mcp add --transport http viable https://api.viablelab.org/mcp

Codex CLI:

codex mcp add viable --url https://api.viablelab.org/mcp

Cursor, in .cursor/mcp.json:

{ "mcpServers": { "viable": { "url": "https://api.viablelab.org/mcp" } } }

VS Code, in .vscode/mcp.json:

{ "servers": { "viable": { "type": "http", "url": "https://api.viablelab.org/mcp" } } }

Another assistant that connects to MCP servers over HTTP takes the same address. Then ask it:

Find the standard for adding fractions with unlike denominators and the misconceptions that get in its way. Then write teaching settings that ask first, and show me the instructions the AI would get.

What it can do

With no key:

• Find a standard. Search K-12 standards by words or code, and read one with its learning components (find_standards, get_standard).
• Find misconceptions. Math misconceptions from Eedi's graph, with the skills they hinder (find_misconceptions).
• Write teaching instructions. Turn teaching settings into the instructions a reply step gets (teaching_instructions).
• Search public goods. The catalog of open datasets, benchmarks and tools for K-12 AI (find_public_goods).
• Check a document. An agent file, a course package or teaching settings, against VIABLE's formats (validate_document).
• List components. What the VIABLE Harness carries (list_components).

With a key:

• Check a reply. Review it with VETTING before a learner sees it (check_reply). The key needs Model requests.
• Run your agents. List your project's published agents and run one (list_published_agents, run_published_agent). The key needs Run agents.

Without a key, no tool calls a model. The standards come from Learning Commons and the misconceptions from Eedi, both under CC BY 4.0: the results carry the attribution to keep.

A reply counts as checked only when check_reply says passed. Your assistant decides whether to call it: nothing makes it check a reply first.

Add a key for the rest

Claude Code:

claude mcp add --transport http viable https://api.viablelab.org/mcp \
  --header "Authorization: Bearer $VIABLE_API_KEY"

Codex CLI:

codex mcp add viable --url https://api.viablelab.org/mcp --bearer-token-env-var VIABLE_API_KEY

Cursor, in .cursor/mcp.json:

{
  "mcpServers": {
    "viable": {
      "url": "https://api.viablelab.org/mcp",
      "headers": { "Authorization": "Bearer ${env:VIABLE_API_KEY}" }
    }
  }
}

VS Code, in .vscode/mcp.json:

{
  "servers": {
    "viable": {
      "type": "http",
      "url": "https://api.viablelab.org/mcp",
      "headers": { "Authorization": "Bearer ${input:viable-key}" }
    }
  },
  "inputs": [{ "id": "viable-key", "type": "promptString", "description": "VIABLE API key", "password": true }]
}

Model calls are billed to the key's project like any other call to the API, and the key's scopes, allowed models and limits apply. Each one appears in Operate → Request logs.

Note

An assistant that only asks for an address can't send a key yet: it gets the tools that need none. Signing in with OAuth, for those assistants, is planned.

What VIABLE keeps

• Request logs keep no tool's arguments or results. A log row holds the time, the status and how long it took, your address, your assistant's user agent and what Cloudflare reports about the connection, the request's headers with credentials removed, and the tool's name. With a key, also the key's prefix and project; for a call that used a model, its models, tokens and cost.
• A run of your agent is kept like any other. run_published_agent runs your agent the way the API does, so its message and reply are in the agent's Test → Run history. Send only what your project may keep.
• Nothing else is kept. The other tools hold what you send only for as long as the call takes. check_reply sends it to the reviewing model, and find_standards sends your words to Learning Commons, whose answers are cached for a day.

Warning

Never send a learner's name or email. Tools that take learners' records aren't on this server at all.

Limits and errors

Where an assistant shows MCP Apps, check_reply's result also shows as a small page in the conversation: the outcome and each review round. Other assistants get the same result as text.