RTK Rewrite

RTK rewrite plugin for OpenClaw. Intercepts exec tool calls and delegates rewrites to rtk rewrite to reduce token usage while preserving command intent.

This plugin appears to do exactly what it claims: intercept exec calls and call your local `rtk rewrite`. Before installing, verify and trust the `rtk` binary you will delegate to (review its source or install method). Avoid enabling audit logging if your commands may contain secrets (or ensure auditDir is a secure local path). Prefer installing RTK from a trusted package manager over piping a remote install script to sh. Test the plugin in a safe environment (verbose mode, with rewriting enabled/disabled) to confirm rewrites are correct and do not change command intent in ways you don't expect.

Type: OpenClaw Skill Name: rtk-rewrite Version: 0.15.3 The plugin intercepts all 'exec' tool calls to rewrite commands via an external 'rtk' binary, which is a high-risk capability as it handles all command strings executed by the agent. While the logic appears aligned with its stated purpose of token optimization, the bundle contains minor anomalies including a future-dated 'publishedAt' timestamp in _meta.json and mismatched repository URLs between SKILL.md and openclaw.plugin.json (rtk-ai/rtk vs aimagist/openrtklaw). Additionally, the documentation encourages a high-risk 'curl | sh' installation method for the required dependency, which is a common vector for supply chain attacks.