TypeScript SDK

The official TypeScript recorder @agent-replay/record produces canonical JSON traces that the Python CLI can import and analyze.

Install

npm install @agent-replay/record

Requires Node.js 22.6+. Zero runtime dependencies.

Quick start

import { trace } from "@agent-replay/record";

await trace(
  { agent: "research-agent", task: "Find pricing", out: "trace.json" },
  async (run) => {
    run.llmCall({
      provider: "anthropic",
      model: "claude-4",
      inputMessages: [{ role: "user", content: "Find pricing" }],
      outputMessage: { role: "assistant", content: "Searching..." },
      usage: { inputTokens: 100, outputTokens: 50 },
    });

    run.toolCall({
      name: "web_search",
      input: { query: "pricing page" },
      output: { url: "https://example.com/pricing" },
    });
  }
);

Then analyze with the Python CLI:

agent-replay import trace.json
agent-replay analyze latest

API

trace(options, fn)

Records a run and writes it to a JSON file.

Options:

• agent (string) — Agent name
• task (string) — Task description
• out (string) — Output file path
• metadata (object, optional) — Arbitrary metadata

run.llmCall(options)

Record a language model call.

run.toolCall(options)

Record a tool invocation.

run.retrieval(options)

Record a retrieval query.

run.memoryRead(options) / run.memoryWrite(options)

Record memory operations.

scrub(data, patterns?)

Redact sensitive fields before recording. Uses DEFAULT_SECRET_PATTERNS (11 patterns) by default.

import { scrub } from "@agent-replay/record";

const clean = scrub({ api_key: "sk-123", data: "safe" });
// { api_key: "[REDACTED]", data: "safe" }