proactive-agent

What This Skill Does

The proactive-agent skill transforms standard AI assistants into autonomous partners by integrating a persistent, self-healing memory architecture. Unlike traditional reactive bots that wait for direct instructions, this agent uses a heartbeat mechanism and reverse prompting to anticipate user requirements. It manages context via pre-compaction flushing, ensuring critical information is preserved even as memory buffers reach their limits. With built-in onboarding flows that can be either interactive or gradual, the agent builds a comprehensive profile (USER.md and SOUL.md) to align its operations with your specific preferences, mission, and long-term objectives.

Installation

To integrate this agent into your OpenClaw workspace, use the following terminal command: clawhub install openclaw/skills/skills/bodii88/proactive-agent-1-2-4 Once installed, place the necessary configuration assets into your project root. The agent will automatically detect the ONBOARDING.md file on the next execution to begin the initialization process.

Use Cases

• Project Management: The agent monitors project files and proactively suggests next steps or flags potential bottlenecks before you encounter them.
• Personal Productivity: By learning your work habits through the 'drip' onboarding mode, the agent creates automated summaries and check-ins at optimal times.
• Agentic Maintenance: Use this for complex, long-running agent workflows where memory decay or prompt-injection security risks are a concern.

Example Prompts

1. "Review my recent project progress and proactively identify three areas where I am at risk of missing a deadline."
2. "I'm feeling overwhelmed by my current task list. Can you suggest a prioritized plan and ask me any questions you need to better understand my goals?"
3. "Run a security audit on my current agent configuration and propose any hardening changes needed based on my recent interactions."

Tips & Limitations

• Onboarding: We recommend the 'Drip' mode for long-term alignment; it allows the agent to learn your nuance through natural conversation rather than a cold interview.
• Memory Management: Ensure the agent has write access to your local directory so it can effectively manage its SOUL.md and memory compaction files.
• Limitations: This skill is designed for agents that require high levels of autonomy. It may not be suitable for static, single-turn tasks where overhead is unnecessary. Always review the 'Reverse Prompting' suggestions before allowing the agent to execute major structural changes.