Claude API (Anthropic SDK)

Use this when you're building a custom agent with the Anthropic SDK and want to pass Vlozi tools to Claude's tool_use. If you just want Vlozi inside Claude Desktop or Claude Code, see the MCP clients page instead — those clients speak MCP natively.

The pattern

Every turn:

Fetch the tool list from GET /tools
Pass it to Claude as tools
When Claude returns a tool_use block, call POST /tools/<name> and feed the result back as tool_result

Example (TypeScript)

import Anthropic from "@anthropic-ai/sdk";
 
const VLOZI_KEY = process.env.VLOZI_API_KEY!;
const MCP = "https://mcp.vlozi.app";
const anthropic = new Anthropic();
 
async function listVloziTools() {
  const r = await fetch(`${MCP}/tools`, {
    headers: { Authorization: `Bearer ${VLOZI_KEY}` },
  });
  const { data } = await r.json();
  // Claude doesn't allow "." in tool names — map blog.list_posts ↔ blog_list_posts
  return data.tools.map((t: any) => ({
    name: t.name.replace(".", "_"),
    description: t.description,
    input_schema: t.inputSchema,
  }));
}
 
async function callVloziTool(name: string, input: Record<string, unknown>) {
  const r = await fetch(`${MCP}/tools/${name.replace("_", ".")}`, {
    method: "POST",
    headers: {
      Authorization: `Bearer ${VLOZI_KEY}`,
      "content-type": "application/json",
      "x-agent-id": "my-custom-agent",
    },
    body: JSON.stringify(input),
  });
  return r.json();
}
 
async function runAgent(prompt: string) {
  const tools = await listVloziTools();
  const messages: Anthropic.MessageParam[] = [{ role: "user", content: prompt }];
 
  while (true) {
    const res = await anthropic.messages.create({
      model: "claude-sonnet-4-6",
      max_tokens: 4096,
      tools,
      messages,
    });
 
    if (res.stop_reason !== "tool_use") return res.content;
 
    const toolResults: Anthropic.ToolResultBlockParam[] = [];
    for (const block of res.content) {
      if (block.type === "tool_use") {
        const result = await callVloziTool(block.name, block.input as any);
        toolResults.push({
          type: "tool_result",
          tool_use_id: block.id,
          content: JSON.stringify(result),
        });
      }
    }
    messages.push({ role: "assistant", content: res.content });
    messages.push({ role: "user", content: toolResults });
  }
}
 
await runAgent("Show my last 3 published posts and write a LinkedIn summary.");

The Python pattern is identical — replace SDK imports with anthropic.Anthropic(); the structures map 1-to-1.

Tips

Tool name dots: Claude's tool_use doesn't allow . in names. Map blog.list_posts ↔ blog_list_posts.
Filter tools per turn: Don't pass all 14 every call — Claude picks better with 3–6 relevant tools.
Set x-agent-id: A stable per-agent string ("ops-bot", "content-bot") shows up in your audit log.
Cache the tool list: Stable for the lifetime of an API key's scopes. Cache for an hour.