Usewhisper Autohook

What This Skill Does

The usewhisper-autohook skill provides a seamless bridge between your OpenClaw agent and the Whisper Context API. It acts as an automated memory management layer, ensuring your agent maintains long-term continuity across sessions. By automatically injecting relevant historical context before generation and ingesting new interactions afterward, the skill eliminates the need for manual memory handling. It supports delta compression and caching, significantly reducing token consumption while maintaining the agent's ability to recall past discussions, user preferences, and project-specific details. Designed specifically for stable session-based platforms like Telegram, it handles state persistence automatically.

Installation

To install this skill, use the ClawHub command-line interface. Ensure you have Node.js installed on your system as it is a required dependency. Run the following command in your terminal:

npx clawhub@latest install usewhisper-autohook

After installation, you must configure your environment variables. OpenClaw relies on these to authenticate with the Whisper Context service. Ensure WHISPER_CONTEXT_API_KEY and WHISPER_CONTEXT_PROJECT are set in your environment configuration. If you are using a self-hosted instance or a custom API endpoint, you may optionally configure WHISPER_CONTEXT_API_URL.

Use Cases

• Personalized Assistants: Remember specific user preferences, hobbies, or past anecdotes across weeks of interaction without cluttering your system prompt.
• Project Tracking: Maintain a persistent knowledge base for professional agents tasked with managing complex, multi-turn coding or writing projects.
• Customer Support Agents: Recall previous tickets or common issues specific to a user ID, providing a more consistent and helpful support experience.

Example Prompts

1. "Hey, remember how I told you last week that I prefer my code documentation to be concise? Apply that same style to this new module."
2. "Based on our discussion about the project architecture yesterday, how does this new feature request fit into our existing schema?"
3. "Summarize what we have decided so far about the marketing strategy for the Q4 launch."

Tips & Limitations

For optimal performance, always ensure your user_id and session_id are stable identifiers. If these values change frequently, the skill cannot effectively map context to the correct user. If you find your agent is still suffering from high token usage due to full chat history replays, consider utilizing the built-in OpenAI-compatible proxy mode to intercept and prune payloads before they reach the model provider.