prompt

Free

System prompt for designing AI agent cost observability and budget governance

FreeFree tier

Type

Open Source

Visit Website GitHub (40 stars)

About prompt

This is a system prompt for an Agent Cost Observability Architect, designed to be used with large language models. The prompt instructs the AI to act as an architect responsible for designing and implementing an end-to-end cost-observability and budget-governance system for AI coding agents such as Claude Code, Codex CLI, Cursor, OpenCode, Gemini CLI, Roo Code, Zed Agent, Goose, GitHub Copilot, and similar tools. The system must normalize multi-provider token pricing into a single cost-per-action metric, build real-time dashboards (including a TUI), enforce per-project budget envelopes, and surface optimization opportunities without blocking developer velocity. The prompt includes details on designing a provider-pricing registry, instrumenting agent sessions, supporting push and pull telemetry patterns, and building anomaly detection for cost spikes.

Key Features

Multi-provider token telemetry normalization into a single cost-per-action metric

Real-time cost dashboards including TUI for current-session burn and rolling totals

Budget governance with per-project envelope enforcement and anomaly detection

Provider-pricing registry that auto-updates from published rate cards with versioning

Support for push (agent-side hook) and pull (proxy/interceptor) telemetry patterns

Instrumentation of every agent session with structured cost events

Pros & Cons

Pros

Provides real-time visibility into previously invisible token costs
Enables proactive budget governance and anomaly detection
Supports multiple AI providers and normalizes pricing differences
Designed to work with existing agents via push/pull telemetry without code changes for legacy agents

Best For

Monitoring token costs across multiple AI coding agents (Claude Code, Codex, Cursor, etc.) in real timeEnforcing per-project budget envelopes to prevent cost overrunsNormalizing and comparing costs across different AI providersOptimizing AI agent usage by identifying expensive operations and optimization opportunitiesProviding cost visibility to developers and managers without blocking velocity

FAQ

What is the purpose of this prompt?

The prompt instructs an AI to act as an Agent Cost Observability Architect, responsible for designing and implementing an end-to-end cost-observability and budget-governance system for AI coding agents across multiple projects, teams, and harnesses.

Which AI coding agents does this prompt cover?

According to the prompt, it covers Claude Code, Codex CLI, Cursor, OpenCode, Gemini CLI, Roo Code, Zed Agent, Goose, GitHub Copilot, and similar tools.

What are the core responsibilities outlined in the prompt?

The core responsibilities include designing multi-provider token telemetry normalization, building real-time cost dashboards (including a TUI), enforcing budgets before they burst, and surfacing optimization opportunities without blocking developer velocity.