An advanced architectural framework that transforms AI agents from reactive followers into proactive partners with persistent memory and self-healing capabilities.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install proactive-agent-3-1-0
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install proactive-agent-3-1-0 using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Proactive Agent?
The Proactive Agent is a comprehensive set of battle-tested patterns designed to give AI agents a sense of continuity, initiative, and self-improvement. By moving beyond simple prompt-response cycles, this skill implements a sophisticated memory architecture including the WAL (Write-Ahead Logging) Protocol and a Working Buffer to survive context truncation. It is a core component of the Hal Stack, designed to ensure that your agent doesn't just wait for instructions but actively anticipates needs and creates leverage through reverse prompting.
Built for high-stakes environments where context loss is not an option, this framework provides the tools for agents to think like owners. It includes robust security hardening to prevent context leakage and unauthorized command execution, making it one of the most resilient Openclaw Skills available. Whether managing complex development workflows or serving as a personalized executive assistant, the Proactive Agent focuses on compounding value through stable, explainable evolution.
Proactive Agent Use Cases
- Managing long-term projects where maintaining state across multiple sessions is critical.
- Automating background maintenance and system checks using autonomous cron jobs that don't interrupt the main thread.
- Onboarding to new workspaces by automatically generating identity and preference files based on user interaction.
- Recovering conversation context and critical decisions immediately after a context window flush or compaction.
- Implementing self-healing workflows where the agent autonomously researches and fixes its own errors before reporting back.
How Proactive Agent Works
- Initial Onboarding: The agent identifies ONBOARDING.md in the workspace and initiates a discovery phase to populate USER.md and SOUL.md.
- WAL Execution: For every user message containing a decision, preference, or correction, the agent updates SESSION-STATE.md before generating a response to ensure the detail is persisted.
- Context Monitoring: When the session context reaches 60% capacity, the Working Buffer Protocol activates, logging all subsequent exchanges to a dedicated file to safeguard against truncation.
- Autonomous Tasking: Background tasks are routed through isolated agent turns rather than system events, allowing the agent to perform maintenance without user intervention.
- Continuous Distillation: The agent periodically reviews daily logs to distill long-term wisdom into the MEMORY.md file, ensuring the most important lessons remain accessible.
- Verification: Before reporting any task as done, the agent follows the Verify Before Reporting (VBR) protocol to test the actual implementation, not just the intent.
Proactive Agent Setup
To integrate this framework into your Openclaw Skills workspace, follow these steps:
- Copy the necessary markdown assets to your project root:
cp assets/*.md ./
- Interaction: Once the files are present, the agent will detect ONBOARDING.md and begin the setup process.
- Security Audit: Run the included security script to verify the integrity of the environment:
./scripts/security-audit.sh
Proactive Agent Data Schema & Taxonomy
The Proactive Agent organizes its intelligence across a specific file hierarchy to ensure persistence:
| File | Purpose | Update Frequency |
|---|---|---|
SESSION-STATE.md |
Active working memory and task state | Every critical detail |
SOUL.md |
Core identity, principles, and boundaries | Infrequent |
USER.md |
User goals, preferences, and context | Continuous |
MEMORY.md |
Curated long-term wisdom and distilled logs | Periodic |
memory/working-buffer.md |
Raw exchange logs for danger-zone survival | High (when >60% context) |
HEARTBEAT.md |
Self-improvement and security checklist | Scheduled |
Proactive Agent Advanced Features
- WAL (Write-Ahead Logging) Protocol that mandates state updates prior to any response to prevent context-loss-induced amnesia.
- Relentless Resourcefulness directive requiring the agent to attempt 10 different technical approaches before requesting human assistance.
- ADL (Anti-Drift Limits) and VFM (Value-First Modification) protocols to ensure the agent's self-evolution remains stable and high-value.
- Compaction Recovery sequence that allows the agent to autonomously rebuild its mental model after a session reset by reading the Working Buffer.
- Unified Search Protocol utilizing semantic search and grep fallbacks across all Openclaw Skills documentation and local logs.
SKILL.md
Loading
Related Openclaw Skills
Spawn Incubator
A decentralized incubator protocol on Base L2 designed to fund, launch, and graduate AI agents into independent businesses.
São Paulo NFS-e Billing & Invoice Management
A specialized skill for automating the issuance, cancellation, and reporting of São Paulo Service Invoices (NFS-e) directly through your AI agent.
Bookmark Intelligence
An automated AI agent that transforms X bookmarks into actionable insights and relates them to your specific development projects.
Amplifier Multi-Agent Delegation
A specialized multi-agent framework that empowers AI agents to delegate complex, high-effort tasks to a fleet of specialized sub-agents.
AI Review Assistant
A strategic AI analyst skill that transforms URLs, PDFs, and media into structured, analytical Markdown reviews.
4todo
A task management skill that integrates the 4todo Eisenhower Matrix into AI agent workflows for seamless prioritization.
Featured*
