tokencast

Tokencast provides pre-execution cost estimation for LLM agent workflows, letting users get cost estimates before initiating tasks and automatically learn from real usage data to improve estimate accuracy. It is designed for developers and teams working with LLM agent pipelines, especially those using IDEs like Cursor, VS Code (with Copilot), Windsurf, and Claude Code. Tokencast works as an MCP server, exposes multiple tools for estimating and reporting costs, supports automatic calibration based on actuals, and integrates easily via plugins or Python API.