rtk-ai/rtk

Feature Request: Support OpenClaw CLI commands for token optimization

Open

#1,244 opened on Apr 12, 2026

View on GitHub
 (1 comment) (2 reactions) (0 assignees)Rust (2,914 forks)batch import
area:clieffort-mediumenhancementhelp wantedpriority:low

Repository metrics

Stars
 (48,085 stars)
PR merge metrics
 (Avg merge 11d 1h) (45 merged PRs in 30d)

Description

Summary

RTK is great for standard dev commands, but it doesn't rewrite OpenClaw CLI commands, which are heavily used in OpenClaw-based AI agent sessions (10-20+ calls per session).

Frequently used OpenClaw commands that need RTK coverage

# Most frequent - sent every response via Telegram/Discord
openclaw message send --channel telegram --target <ID> --message ...
openclaw message send --channel discord --message ...

# Status/debug
openclaw status
openclaw gateway status
openclaw plugins list

# Model management
openclaw models status

# Cron
openclaw cron list

# Wiki
openclaw wiki search <query>
openclaw wiki ingest <file>

Suggested token savings

  • openclaw message send output: strip plugin loading logs ([plugins] ..., [lcm] ...) - saves ~100-200 tokens per call
  • openclaw status output: compact format with key metrics only
  • openclaw plugins list: table → compact list
  • openclaw cron list: table → compact list

Current behavior

rtk rewrite returns exit code 1 (no match) for all openclaw commands:

$ rtk rewrite "openclaw message send --channel telegram --target 6572517363 --message test"
# no output, exit 1

Environment

Impact estimate

In a typical OpenClaw session, openclaw message send is called 10-20 times. Each call outputs ~150 tokens of plugin loading logs that are injected into LLM context. Adding RTK support could save 1,500-3,000 tokens per session just from message sends alone.

Thank you for the great tool! 🦞

Contributor guide