Show HN: Track Token Usage from Codex CLI with Splitrail (OSS, Rust) (github.com)

Splitrail is an open-source, Rust-built CLI tool that tracks token usage and costs in real time for coding-focused LLMs (Gemini CLI, Anthropic Claude Code and OpenAI Codex). Run a single command to see live token counts and cost estimates for your CLI coding agents, and optionally upload encrypted usage to a private Splitrail Cloud account to aggregate data across machines. Binaries are available on the Releases page, or you can build from source with Cargo; Windows builds require LLVM’s lld-link (installable via winget) added to PATH to speed linking. Why it matters: as more developers embed code-writing agents in workflows, fine-grained, low-latency token accounting becomes crucial for cost control, auditing, and reproducibility. Splitrail’s Rust implementation aims for performance and portability, making it suitable for local dev machines, CI, or multi-machine teams. The project also ties into Piebald, the team’s developer-first agentic AI experience. Caveats: cloud upload enables cross-machine aggregation but introduces a surface for privacy considerations, so teams should weigh local-only versus cloud-backed tracking. Overall, Splitrail fills a practical gap for teams needing transparent, realtime visibility into coding-agent consumption.