@piiiico/agent-logger

Zero-config tool call logger for AI agents. One import, full audit trail. Works with Anthropic/Claude, OpenAI, LangChain, and CrewAI.

Zero dependencies. Logs to stdout, file, or AgentLair cloud for search, compliance, and governance.

Install

npm install @piiiico/agent-logger
# or
yarn add @piiiico/agent-logger
# or
bun add @piiiico/agent-logger

Quick start

import { auditLog } from "@piiiico/agent-logger";

await auditLog({
  agent: "customer-support-bot",
  action: "respond",
  tool: "knowledge_base_search",
  input: { query: "How do I cancel my subscription?" },
  output: { answer: "Go to Settings → Billing → Cancel." },
});
// → {"agent":"customer-support-bot","action":"respond","tool":"knowledge_base_search","input":{...},"output":{...},"timestamp":"2026-04-01T12:00:00.000Z"}

By default, logs go to stdout as JSON lines — pipe them wherever you want.

Core API

auditLog(entry: AuditEntry, opts?: AuditLogOptions): Promise<AuditResult>

AuditEntry fields

| Field | Type | Required | Description | |---|---|---|---| | agent | string | ✅ | Agent identifier | | action | string | ✅ | What happened ("llm_call", "tool_use", etc.) | | tool | string | — | Tool or function name | | input | unknown | — | Request payload | | output | unknown | — | Response payload | | timestamp | string | — | ISO 8601 — defaults to Date.now() | | metadata | Record<string, unknown> | — | Any extra context |

Transports

| Transport | When used | Config | |---|---|---| | "console" (default) | Always — great for dev + log aggregators | — | | "file" | Append to a local file | filePath: "./audit.log" | | "agentlair" | Ship to AgentLair for governed audit trail | Set AGENTLAIR_API_KEY | | "silent" | Testing | — |

Auto-detection: if AGENTLAIR_API_KEY is set, the default transport becomes "agentlair". Otherwise, "console".

Pre-bound logger

import { createAuditLogger } from "@piiiico/agent-logger";

const log = createAuditLogger("inventory-agent");

await log({ action: "check_stock", tool: "db_query", input: { sku: "ABC-123" } });
await log({ action: "reorder", output: { orderId: "PO-9999" } });

Framework adapters

LangChain

Drop AgentAuditCallback into any chain, agent, or model's callbacks array:

import { AgentAuditCallback } from "@piiiico/agent-logger/langchain";
import { LLMChain } from "langchain/chains";
import { ChatOpenAI } from "langchain/chat_models/openai";

const llm = new ChatOpenAI();
const chain = new LLMChain({
  llm,
  prompt,
  callbacks: [new AgentAuditCallback("my-langchain-agent")],
});

await chain.call({ question: "What is 2+2?" });
// Automatically logs: llm_start, llm_end, tool_start, tool_end, chain_start, chain_end

With AgentLair backend:

new AgentAuditCallback("my-agent", { transport: "agentlair" })

Anthropic / Claude SDK

Wrap your Anthropic client — zero behavior change, full logging:

import Anthropic from "@anthropic-ai/sdk";
import { wrapAnthropicClient } from "@piiiico/agent-logger/anthropic";

const client = wrapAnthropicClient(new Anthropic(), "my-claude-agent");

// Identical to normal usage — requests and responses are automatically logged
const message = await client.messages.create({
  model: "claude-3-5-sonnet-20241022",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello!" }],
});

Logs: messages_create (with model + input) and messages_response (with content + token usage).

OpenAI

Wrap your OpenAI client — same pattern, automatic logging:

import OpenAI from "openai";
import { wrapOpenAIClient } from "@piiiico/agent-logger/openai";

const client = wrapOpenAIClient(new OpenAI(), "my-openai-agent");

// Use exactly as before — all calls are logged automatically
const completion = await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Hello!" }],
});

Logs: chat_completions_create (with model + input) and chat_completions_response (with content + token usage + tool calls).

Tool calls in the response are captured automatically:

// If the model calls tools, they're logged with name + arguments
const result = await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Search for AI news" }],
  tools: [{ type: "function", function: { name: "web_search", description: "Search" } }],
});
// → logs tool_calls: [{name: "web_search", arguments: "..."}]

CrewAI

Option A — wrap a task function (JS/TS):

import { wrapCrewTask } from "@piiiico/agent-logger/crewai";

const webSearch = async (query: string) => {
  // ... your search logic
  return results;
};

const auditedSearch = wrapCrewTask(webSearch, {
  agent: "research-crew",
  taskName: "web_search",
});

const results = await auditedSearch("latest AI agent frameworks");
// Logs: crew_task_start → crew_task_end (with duration)

Option B — bridge from Python CrewAI via direct log call:

import { crewAiLog } from "@piiiico/agent-logger/crewai";

// Call this from your JS bridge/API when a Python crew event fires
await crewAiLog({
  agent: "python-research-crew",
  task: "summarize",
  input: { text: "..." },
  output: { summary: "..." },
});

Ship to AgentLair

AgentLair provides an agent governance layer — audit trails, approval gates, and replay for AI agents.

Set your API key and all logs automatically route to your AgentLair dashboard:

AGENTLAIR_API_KEY=your_key_here node your-agent.js

To get access: contact us at [email protected] or open a GitHub issue — we're onboarding teams individually during early access.

File transport example

await auditLog(
  { agent: "batch-processor", action: "process", input: { jobId: "job-42" } },
  { transport: "file", filePath: "/var/log/agents/audit.log" }
);

Appends JSON lines. Compatible with any log aggregator (Datadog, Loki, CloudWatch).

License

MIT