TokenRanger

Install, configure, and operate the TokenRanger OpenClaw plugin. Use when you want to reduce cloud LLM token costs by 50-80% via local Ollama context compres...

This skill is internally consistent with a local compression sidecar, but before installing you should: (1) verify the plugin package and repo (openclaw-plugin-tokenranger) on GitHub and npm to ensure you trust the maintainer; (2) inspect the plugin's service unit / launchd plist and start scripts that will be installed (~/.openclaw/extensions/... and user-level systemd/LaunchAgents) so you know what commands will run as a persistent service; (3) be aware the setup downloads Ollama models and Python deps (large network activity and disk usage) and will run a local HTTP sidecar bound to localhost — confirm it does not expose data to the network; (4) backup any important OpenClaw config before changing plugin entries; and (5) if you need higher assurance, review the plugin source code or test in an isolated VM/container first.

Type: OpenClaw Skill Name: tokenranger Version: 1.0.0 The OpenClaw skill 'tokenranger' is classified as benign. The `SKILL.md` file transparently describes the installation and operation of a local sidecar plugin designed to reduce LLM token costs. It uses standard OpenClaw CLI commands for plugin management and configuration, and standard system commands (`systemctl`, `launchctl`) for registering and managing a user-level background service. While installing a system service is a powerful capability, it is explicitly stated and appears necessary for the plugin's continuous operation as a 'sidecar'. There is no evidence of prompt injection, data exfiltration, unauthorized remote access, or other malicious intent within the provided files. All described actions are consistent with the stated purpose and use local resources (localhost:8100, localhost:11434).