AgentTrace — AI Agent observability. One command to monitor your agents locally.
The open-source circuit breaker and accountability layer for multi-agent AI pipelines. Trace every action. Explain every decision. Short-circuit before the damage propagates.
AgentTrace proxy server for buffering and forwarding agent events
AgentTrace local server with Express and SQLite
AgentTrace SDK for monitoring AI agent activity
Shared types, schemas, and utilities for AgentTrace packages
LangGraph JS adapter for agenttrace-react. Visualise LangGraph agent execution traces, tool calls, and human-in-the-loop interrupts as a live React UI.
AG-UI protocol adapter for agenttrace-react. Visualise AG-UI agent execution traces, tool calls, and human-in-the-loop approval gates as a live React UI.
Vercel AI SDK adapter for agenttrace-react. Visualise AI SDK agent execution traces, tool calls, and human-in-the-loop approval gates as a live React UI.
SDK, CLI, and Claude Code parser for AgentTrace — see where your tokens go
Decision tree data model, waste analyzer, and pricing DB for AgentTrace
The official Language-Agnostic Node.js SDK for AgentTrace. Zero-config visual debugging for LLM agents.
Headless React primitives for AI agent execution traces, human-in-the-loop approval gates, and agent status UI. Works with LangGraph, AG-UI, and the Vercel AI SDK.
Testing utilities, mock providers, fixtures, and eval framework for ElsiumAI
Lightweight observability SDK for AI agents — zero manual tracking
Visual debugger for AI agent loops. Step-by-step. Locally. Zero config.
Signed, hash-chained proof logs for AI agent tool executions and auditable events. MCP-native. Local-first.
See what your AI coding agent actually did — and whether to trust it. A local-first, cross-agent replay + trust report you attach to a PR. Review the process, not just the diff.
Cost + latency aggregation for LLM agent runs. Group calls into named runs, get totals, p50/p95, and per-model breakdowns. Composes with cachebench.
WHERE-layer structured citations for AI agent outputs: attach, validate, and audit source pointers.
WHY-layer decision log for AI agents: record options considered, the option chosen, the rationale, and the outcome, then persist as JSONL.
Tiny in-process pub/sub for agent loop events. Sync-only Rust mirror of the Python agent-event-bus library: register handlers by event name (or `*` for every event), emit, and remove via opaque Subscription handles. Subscriber panics are isolated so the bus keeps dispatching.