nickelsec/bough

Bough is a Go binary that visualizes and calculates the cost of prompts generated by Claude Code, Codex, and Pi. It runs locally, processing session data to display daily work and financial impact. The tool does not transmit data externally, offering a privacy-focused way to audit AI coding agent usage and expenses.

Cover image for nickelsec/bough

Bough operates as a standalone Go binary that converts recorded AI coding agent activity into a visual and financial summary. The software identifies the specific prompts used with Claude Code, Codex, or Pi to display daily workloads. Each recorded interaction is assigned a price to show the total financial impact of the session. The developers state that the tool processes all data locally without transmitting any information to external servers. A user runs the compiled binary on a personal machine to access historical session data from supported coding agents. The program generates a visual representation of the day’s recorded prompts and calculates the associated costs for each entry. This process provides a detailed audit of how specific AI interactions contributed to the overall daily workload. The interface displays the relationship between the generated code or responses and their respective price tags. The official repository notes a count of 101 stars and 11 forks, which serves as an indicator of community engagement. First observed crossing Wirecopy's new-project discovery gate today, the project appears recently arrived in the discovery pool. The description claims that the tool sends nothing anywhere, a privacy assertion that requires verification for strict compliance needs. The accuracy of the cost calculations depends on the pricing structures defined for the specific AI models used in each session.

README

nickelsec/bough View on GitHub

Loading the README from GitHub…