Debuly

agenttrace

Local-first Rust terminal app that audits AI coding agent sessions for cost, tokens, latency and failures from logs on disk.

Open source
Visit github.com agenttrace (official site)Compare developer platforms

What is agenttrace?

agenttrace is an open-source terminal interface and command-line tool written in Rust. It reads the session history that AI coding agents already store on your machine and turns it into views of cost, token use, latency, failures and overall session health. Supported sources include OpenCode, Cursor exports, Aider, Cline, Qwen Code, Codex CLI and Claude Code, among others, so one tool can compare usage across several agents. It helps you spot slow runs and see where time went, and it can generate reports for review. Because it is local-first, no cloud account is needed and your logs stay on your computer. The project is MIT licensed and free, with documentation in English and Chinese. It suits developers and engineering leads who want to understand and control what their coding agents spend.

Use cases

  • Audit coding agent session costs
  • Compare token use across agents
  • Diagnose slow agent runs
  • Find failed agent sessions
  • Generate usage reports
  • Review session health trends

Who it's for

  • Developers using coding agents
  • Engineering leads
  • AI tool power users

Why use agenttrace

  • Agent spending becomes visible, so unexpected cost is easier to catch early.
  • One tool covers several coding agents instead of a separate dashboard for each.
  • Session data stays on your machine because nothing is uploaded.
  • Seeing where time and failures cluster helps you tune your agent workflows.

agenttrace features

  • Terminal UI for browsing session history
  • Cost and token breakdowns per session
  • Latency and slow-run diagnosis
  • Failure and session health tracking
  • Reads logs from Claude Code, Codex CLI, Cline and Aider
  • Handles Qwen Code, OpenCode and Cursor exports
  • Report generation
  • Local-first design with no cloud service

agenttrace integrations

  • Claude Code
  • Codex CLI
  • Qwen Code
  • Cline
  • Aider
  • Cursor
  • OpenCode

How to get started with agenttrace

  1. Install agenttrace with Homebrew or npm, or build it from the Rust source.
  2. Run it so it finds session logs from the coding agents on your machine.
  3. Browse sessions in the terminal interface to see cost, tokens and timing.
  4. Open the health and failure views to find problem sessions.
  5. Generate a report to keep or share for review.

agenttrace limitations

  • Terminal and command-line interface only
  • Accuracy depends on what each agent writes to its session logs

agenttrace pricing

agenttrace is open source, so you can use and self-host it for free.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

agenttrace alternatives

The main agenttrace alternatives listed on Debuly are Langfuse, Helicone, Boost, BurnRate, ClaudeScope and agentmemory. The table compares pricing and platforms; order is A–Z.

agenttrace compared with its alternatives
ToolPricingStarting pricePlatforms
agenttraceOpen source—CLI, Self-hosted
agentmemoryOpen source—CLI, Self-hosted, macOS, Linux
BoostFree—CLI, macOS, Windows, Linux
BurnRateFreemium$12/moWeb, CLI, macOS, Linux
ClaudeScopeOpen source—CLI, Web, macOS, Linux
HeliconeFreemium$79/moWeb, API, Self-hosted
LangfuseFreemium$29/moWeb, API, Self-hosted, Docker

  1. agentmemory

    Open-source memory server that lets AI coding agents like Claude Code and Cursor recall past sessions, with local-first search.

    Developer platformsCodingProductivity
    Open sourceVisit agentmemory (official site)

  2. Boost

    Free JFrog CLI that trims noisy command output before it reaches coding agents like Cursor, Claude Code and Codex, saving tokens and context.

    Developer platformsCoding
    FreeVisit Boost (official site)

  3. BurnRate

    AI cost analytics that combines OpenAI and Anthropic spend in one dashboard, with budgets and a local CLI for coding-tool sessions.

    Developer platformsProductivity
    FreemiumVisit BurnRate (official site)

  4. ClaudeScope

    Free, open-source local dashboard that searches every Claude Code session and shows token, cost and cache-savings usage.

    Developer platformsCodingProductivity
    Open sourceVisit ClaudeScope (official site)

  5. Helicone

    Open-source LLM observability and AI gateway that logs requests, tracks cost and latency, and routes to 100+ models through one API.

    Developer platformsCoding
    FreemiumVisit Helicone (official site)

  6. Langfuse

    Open-source LLM observability and evaluation platform for tracing agents, managing prompts and scoring outputs.

    Developer platformsAutomation and agentsCoding
    FreemiumVisit Langfuse (official site)

agenttrace FAQ

How much does agenttrace cost?

Nothing. It is open source under the MIT license and has no paid plans.

Is there a free plan for agenttrace?

There are no tiers. The whole tool is free to install and run on your own machine.

Which agents does agenttrace support?

Its README names OpenCode, Cursor exports, Aider, Cline, Qwen Code, Codex CLI and Claude Code, among others.

Does agenttrace upload my sessions?

No. It is local-first and works on session history stored on your own computer.

Is agenttrace open source?

Yes. The code is public on GitHub under the MIT license.

What are alternatives to agenttrace?

Alternatives include ccusage, Langfuse, Helicone and other observability tools for LLM and agent usage.

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

SourceTypeAccessed
github.com/luoyuctl/agenttraceofficial

Image credits

  • Logo: none collected yet, so a monogram is shown.
  • Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.