OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Evaluation & Observability

Tokenjam

Tokenjam is a Python-based developer tool for analyzing AI-agent telemetry, identifying excessive token usage and reviewing efficiency improvements through a local browser dashboard.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Tokenjam — OpenViST project profile in Agent Evaluation & Observability.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Tokenjam when you have agent telemetry to inspect and want a local workflow for finding token overspending and reviewing possible efficiency changes.

Who it is for

  • Developers using Claude Code or Codex
  • Developers building agents with Python SDKs or APIs
  • Teams using supported agent frameworks or OpenTelemetry
  • Developers reviewing token consumption and agent cost across workflows

Documented uses

  • Capture or import agent telemetry
  • Compare token and spend patterns across models, sessions or workflows
  • Identify excessive token usage
  • Review proposed fixes and optimization rules
  • Analyze custom SDK- or API-based agents locally
Before you choose

Requirements & limits

The repository is public and MIT-licensed. The documented workflow runs locally on the user's machine and describes no cloud or signup requirement.

  • Whether your agent produces telemetry through one of the documented paths
  • Whether local browser analysis and command-line review fit your workflow
  • Whether you need Claude Code or Codex history backfill, SDK instrumentation, framework patching or OpenTelemetry ingestion
  • Whether custom-agent integration can meet the telemetry requirements, which are not specified in the available description
  • Whether proposed optimization rules are useful for the files and workflows you maintain

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up local telemetry capture and the Tokenjam dashboard

Configure Tokenjam for a supported agent path and open its local analysis dashboard.

Before you start

  • A supported path such as Claude Code, Codex, a Python SDK or API agent, a supported framework, Langfuse or Helicone traces, or an OpenTelemetry-emitting agent.
  • For the command-line setup, use either npx tokenjam or pipx with the tokenjam package.
  • Custom SDK or API integrations require adding @watch() or configuring an OTLP exporter as documented.

Steps

  1. Run npx tokenjam onboard, or install the package with pipx and run tj onboard.

  2. Answer the onboarding questions so Tokenjam can configure the relevant capture path and analyzers.

  3. For SDK or API agents, add @watch() to the code or point an OTLP exporter at the local tj serve endpoint; framework users use the documented patch function path.

  4. Run tj serve and open the local Lens dashboard at http://127.0.0.1:7391/.

  5. Run tj optimize to review costs and proposed fixes, then tj rules list to inspect the resulting rules.

Expected outcome

A local dashboard showing available spend, models, sessions, token use and proposed optimization fixes.

Guide limitations

  • The supplied documentation does not specify the telemetry schema or integration steps for custom SDK or API agents.
  • The plugin does not install the tj command; it must already be available on the PATH.
  • tj rules list reads the latest analysis, so it follows tj optimize or the daemon's first analysis cycle.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
30 September 2026
Original website title
Metabuilder-Labs/tokenjam · Official homepage title · retained in its source language
How we review sourcesSuggest a correction