A sophisticated, self-improving architecture that turns AI agents into proactive partners who anticipate needs and maintain state through context loss.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install alexey-proactive-agent
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 alexey-proactive-agent 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 high-performance framework designed to move AI agents beyond simple task-following. By integrating this into your workflow from the Openclaw Skills library, you enable your agent to think like an owner rather than an employee. It utilizes a three-tier memory system and rigorous state-tracking protocols to ensure that no detail is lost, even during context compaction or session resets.
This skill is part of the Hal Stack and focuses on three core pillars: proactivity, persistence, and self-improvement. It empowers agents to perform reverse prompting, surface valuable ideas unprompted, and employ relentless resourcefulness by attempting multiple technical approaches before seeking human intervention. It is the definitive solution for developers looking for Openclaw Skills that prioritize long-term autonomy and behavioral integrity.
Proactive Agent Use Cases
- Managing complex, long-term projects where context continuity is critical.
- Automating background maintenance and self-healing tasks using autonomous crons.
- Building a personalized AI assistant that adapts to user preferences through a persistent SOUL.md and USER.md.
- Recovering work seamlessly after AI context windows are truncated or flushed.
How Proactive Agent Works
- Input Scanning: The agent constantly monitors user input for specific triggers like corrections, decisions, or proper nouns.
- WAL Protocol Execution: Before generating a response, the agent writes critical details to SESSION-STATE.md to ensure a Write-Ahead Log of state changes.
- Buffer Management: When context usage hits 60%, the agent activates the Working Buffer Protocol to log every exchange in the 'danger zone'.
- Compaction Recovery: Upon detecting a context reset, the agent reads the working buffer and daily logs to reconstruct the active task state.
- Self-Improvement Loops: The agent periodically runs heartbeat checks to distill logs into long-term memory and optimize its own operating rules.
Proactive Agent Setup
To get started with this selection from Openclaw Skills, follow these steps:
# Copy the core asset templates to your workspace
cp assets/*.md ./
# Initialize the agent onboarding
# The agent will detect ONBOARDING.md and ask setup questions
# Run the mandatory security audit script
./scripts/security-audit.sh
Proactive Agent Data Schema & Taxonomy
The Proactive Agent organizes its workspace into a specific hierarchy to ensure data persistence and easy retrieval:
| File | Purpose | Update Cadence |
|---|---|---|
SESSION-STATE.md |
Active working memory/RAM | Every message with critical details |
SOUL.md |
Identity, principles, and boundaries | Rare (Core evolution) |
USER.md |
User goals, preferences, and context | As learned from interaction |
MEMORY.md |
Curated long-term wisdom | Periodic distillation |
memory/working-buffer.md |
Danger zone logs for recovery | Every exchange >60% context |
Proactive Agent Advanced Features
- WAL Protocol: A strict Write-Ahead Logging system that prioritizes state updates over immediate responses.
- Autonomous Crons: The ability to spawn isolated sub-agents for background maintenance without interrupting the main session.
- Anti-Drift Limits (ADL): Guardrails that prevent the agent from adopting unverified or overly complex behavioral changes.
- Unified Search Protocol: A multi-source fallback system that searches semantic memory, grep patterns, and transcripts before reporting a lack of information.
- Relentless Resourcefulness: A core identity directive to try 5-10 different methods to solve a problem before giving up.
SKILL.md
Loading
Related Openclaw Skills
Marketing Skills
A comprehensive suite of 23 marketing playbooks providing frameworks, checklists, and ready-to-use deliverables for growth and optimization.
Marketing Skills
A powerhouse collection of 23 marketing modules providing actionable frameworks for SEO, CRO, copywriting, and growth experiments.
Brave Search Skill
A lightweight CLI tool for performing headless web searches and extracting clean Markdown content without a browser.
Disclawd
A communication platform that enables AI agents to join servers, message users, and participate in real-time Discord-like conversations.
Sales Skill
An elite sales operations partner for AI agents to track leads, manage pipelines, and automate outreach sequences.
RAG System Builder
A comprehensive framework for building fully offline Retrieval-Augmented Generation systems for local document ingestion and semantic search.
Featured*
