token-budgeter

Enforce per-tier token budgets for Claude executions using tier-specific pricing.

Updated Aug 23, 2026

Search Tags:#agent orchestration #FinOps #token budgeting #cost control #Claude tiers #runaway spend

Ready to run

npx skills add https://github.com/CleanExpo/Pi-Dev-Ops --skill token-budgeter

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: token-budgeter
Source: https://github.com/CleanExpo/Pi-Dev-Ops/tree/main/skills/token-budgeter
Command: npx skills add https://github.com/CleanExpo/Pi-Dev-Ops --skill token-budgeter

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you track and enforce token budgets per Claude tier so your agentic runs don’t exceed cost expectations.

Core Features & Use Cases

Tier-aware budgeting: Applies distinct token cost assumptions to Opus 4.7, Sonnet 4.6, and Haiku 4.5 output (and cache) pricing.
Budget enforcement: Keeps planning/execution aligned with a target spend by monitoring token usage by tier.
Zero-cost Max execution guidance: Clarifies that Claude Max runs as $0 in this workflow because execution is via the local claude CLI subprocess.
Use Case: You’re running repeated autonomous repository scans and want to stop or rebalance the workflow once cumulative token usage threatens your weekly budget.

Quick Start

Tell the agent to enable the token-budgeter for Opus/Sonnet/Haiku and halt planning when the configured per-tier token budget would be exceeded.

Frequently Asked Questions about token-budgeter

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I prevent runaway spend during autonomous Claude agent workflows?▼

Yes, you can apply tiered token budgeting across Opus 4.7, Sonnet 4.6, and Haiku 4.5 by calculating distinct cost assumptions for each tier's output and cache pricing, then halting or rebalancing planning when limits are reached.

How do I set token cost limits for Opus 4.7, Sonnet 4.6, and Haiku 4.5 tiers?▼

You set token cost limits by configuring tier-specific token budgets that estimate output and cache costs using Opus 4.7, Sonnet 4.6, and Haiku 4.5 pricing, then monitoring usage to trigger budget stop or adjust behavior.

Does Claude Max execution count toward the token budgeting limits?▼

Claude Max execution does not count toward token budgeting limits because it runs as $0 cost in this workflow, with execution handled locally via the claude CLI subprocess rather than the metered API tiers.

What is the best way to stop autonomous repository scans from exceeding my weekly budget?▼

The best way to stop autonomous repository scans from exceeding a weekly budget is to enable tier-aware token budgeting that monitors cumulative usage across Opus, Sonnet, and Haiku, triggering a halt when the configured spend limit is reached.

Can I rebalance agentic planning workflows to a cheaper Claude tier when cost limits approach?▼

Yes, you can rebalance agentic planning workflows to a cheaper Claude tier by monitoring token usage by tier against configured budgets, which triggers budget adjust behavior to keep spend within your target limits.

Why do I need tier-specific pricing for FinOps cost control in agent orchestration?▼

You need tier-specific pricing for FinOps cost control in agent orchestration because Opus 4.7, Sonnet 4.6, and Haiku 4.5 each carry distinct output and cache costs, requiring separate budget tracking to prevent overspend.