# I need to track and cut coding-agent token costs

> Local trackers and proxies that show what your coding agents spend on tokens and cut how many they send.

The best fits to track and cut coding-agent token costs on Agent Switchboard, as of 20 Nov 2026: CodeBurn, Tokscale and agentsview, of 12 tools that do it.

## Quick answers

- **Which official tools track and cut coding-agent token costs?** Langfuse.
- **Which have a hosted MCP server?** None of the 12 tools lists a hosted MCP server address.
- **Which need no API key or sign-in?** CodeBurn, Tokscale and agentsview, among 8 that need none.

## Best fits

1. [CodeBurn](https://agentswitchboard.dev/agents/codeburn) (cli): Tracks AI coding token usage and cost across coding agents locally, from a CLI, a menubar app and a web dashboard.
2. [Tokscale](https://agentswitchboard.dev/agents/tokscale) (cli): Tracks token usage and cost across AI coding agents from a terminal UI and a web dashboard, with live pricing.
3. [agentsview](https://agentswitchboard.dev/agents/agentsview) (cli, mcp): Cost breakdown: Calculates token cost per model with prompt-caching-aware pricing, filtered by date or agent.
4. [Claw Compactor](https://agentswitchboard.dev/agents/claw-compactor) (api, cli): Compresses LLM context such as code, JSON, logs, and chat history through a reversible 14-stage pipeline with no LLM calls, via a Python API and CLI, for agent developers.
5. [LiteLLM](https://agentswitchboard.dev/agents/litellm) (api, cli): AI gateway unifying 100+ LLM providers under one OpenAI-compatible API. Cost tracking, guardrails, load balancing. pip install litellm. 47k GitHub stars.
6. [billion-context](https://agentswitchboard.dev/agents/billion-context) (cli): Compresses context in a local proxy between an agent and its model API, so very long sessions keep going, with a Claude Code plugin.
7. [Headroom](https://agentswitchboard.dev/agents/headroom) (api, mcp, cli): Compresses tool outputs, logs, RAG chunks, files, and history locally before they reach the model, as a Python or TypeScript library, a drop-in proxy, an agent wrapper, or an MCP server.
8. [pxpipe](https://agentswitchboard.dev/agents/pxpipe) (cli): Cuts coding agents' input tokens by rendering bulky context as images in a local proxy for Claude Code, Gemini CLI and Antigravity.
9. [RTK](https://agentswitchboard.dev/agents/rtk) (cli): Filters and compresses shell command output such as git, test runs and builds before a coding agent reads it, via hooks for Claude Code, Cursor, Codex and Gemini CLI.
10. [AgentOps](https://agentswitchboard.dev/agents/agentops) (api): Track token usage and costs per agent run across all major LLM providers
11. [Maki](https://agentswitchboard.dev/agents/maki) (cli): Rust coding agent CLI with tree-sitter code indexing, sandboxed Python execution, and multi-tier subagents. Benchmarks show ~40% token reduction vs competing coding agents.
12. [Langfuse](https://agentswitchboard.dev/agents/langfuse) (api, cli · Official): Analyze user session funnels, latency distributions, and token cost trends

## For agents

- This task as JSON: https://agentswitchboard.dev/api/tasks/track-coding-agent-costs
- One pick for your user's need, with the next step: https://agentswitchboard.dev/api/recommend?need=track%20and%20cut%20coding-agent%20token%20costs (or the `recommend` MCP tool at https://agentswitchboard.dev/api/mcp)
- Every task: https://agentswitchboard.dev/tasks · the whole site: https://agentswitchboard.dev/llms.txt

Updated 20 Nov 2026 with the weekly drop. Ranked by how well each listing matches the task, never by payment. Page: https://agentswitchboard.dev/tasks/track-coding-agent-costs
