Session Init

What This Skill Does

The Session Init skill, developed by earnabitmore365, serves as a mission-critical synchronization mechanism for OpenClaw AI agents. It is designed to be triggered at the start of every new session to ensure the agent maintains full contextual awareness and adherence to its core principles. By systematically reviewing cross-session memory, persona configuration files, available skill sets, and MCP (Model Context Protocol) server status, the skill eliminates the 'inertia' often found in long-term AI interactions. It enforces the 'Truth Engine' protocol—a commitment to data integrity where the agent prioritizes transparency over fabrication, ensuring that if information is unavailable, the agent acknowledges it honestly.

Installation

To integrate this skill into your environment, use the OpenClaw command-line interface. Run the following command in your terminal:

clawhub install openclaw/skills/skills/earnabitmore365/session-init

Ensure that you have sufficient permissions for the agent to access its workspace directories, as the skill performs file-read operations on ~/.openclaw/workspace/ and ~/.agents/skills/ to verify current system status.

Use Cases

• Contextual Resumption: Ideal for developers or power users who maintain long-term, complex projects with the AI and need to re-align the agent's focus after a break.
• Persona Enforcement: Use this to force the AI to refresh its core identity (AGENTS.md, SOUL.md, IDENTITY.md) and reset its behavior patterns to align with your pre-defined goals.
• System Audits: Useful for verifying that all MCP servers and custom skill extensions are mounted correctly and ready for use at the start of a deep-work session.
• Error Recovery: Trigger this when the AI feels 'lost' or deviates from established operating parameters to reset the interaction baseline.

Example Prompts

1. "初始化：请开始新会话检查，并回顾最近的记忆碎片。"
2. "init：我们需要进行一次全面的启动检查，确保所有人格文件和 MCP 服务器状态正常。"
3. "session-init：执行人格巩固程序，重新审阅官方和自定义 Skills 列表。"

Tips & Limitations

• Strict Compliance: The skill follows an uncompromising 8-step execution flow. Do not attempt to bypass these steps, as the verification evidence is mandatory for compliance.
• Transparency: Every step requires log outputs (like cat, ls, or mcporter info). The agent will provide these logs to prove it has actually performed the read/check operations.
• Truth Engine: Always review the final Checklist generated by the agent. If any item is marked as incomplete, pause operations to identify the system bottleneck or file access issue.
• Resource Usage: Because this skill performs multiple file system reads, it is best used once per session start rather than repeatedly to save on latency and processing overhead.