Tag
The rtk library saves 2.5M tokens across coding agents in 2 weeks by compacting shell command outputs, reducing token consumption.
GrepSeek trains LLM search agents to directly interact with a text corpus using shell commands like grep, using a two-stage training pipeline with cold-start dataset construction and GRPO refinement, achieving strong F1 and Exact Match on open-domain QA benchmarks.
Tips for using Cron jobs and Heartbeat in OpenClaw to improve efficiency and reduce token usage, with examples of when to use each.