A technical framework for managing long-running agent sessions through structured summarization and tokens-per-task optimization.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install hz-context-optimizer
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 hz-context-optimizer 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 Context Compression?
Context Compression is a sophisticated strategy designed to handle agent sessions that exceed standard token limits. Instead of naive truncation, this skill focuses on tokens-per-task optimization, ensuring that critical information like file modifications and architectural decisions are preserved. By utilizing Openclaw Skills for context management, developers can build more reliable agents that do not lose track of their progress during complex, multi-step coding tasks.
The core value lies in trading raw token volume for information density. It prioritizes structured data over simple character reduction, employing methods like Anchored Iterative Summarization to maintain a persistent record of session intent and file changes. This ensures that even when millions of tokens are processed, the agent maintains a high-fidelity understanding of the system state.
Context Compression Use Cases
- Managing agent sessions that exceed 1M+ tokens in conversation history.
- Maintaining state in long-running coding tasks where multiple files are modified.
- Reducing API costs by optimizing tokens-per-task rather than just tokens-per-request.
- Preventing agent amnesia in large-scale system refactors using Openclaw Skills.
- Debugging sessions where the AI model loses the reasoning chain or technical context.
How Context Compression Works
- The system monitors context window utilization, typically triggering when the session reaches 70-80% capacity.
- It identifies relevant technical metadata, including session intent, modified files, and pending decisions.
- An Anchored Iterative Summarization process is initiated to merge new information into a persistent, structured summary rather than regenerating from scratch.
- The skill applies probe-based evaluation to verify that critical details like file paths and error codes remain intact during the compression cycle.
- The optimized context is then fed back into the agent loop, enabling continued task execution within token constraints.
Context Compression Setup
To integrate context compression within your agent framework using Openclaw Skills, configure your environment with the following parameters:
npm install @openclaw/context-manager
# Or configure via your agent's skill manifest
openclaw-cli add context-compression
Define your summary anchors in a configuration file to ensure sections like 'Files Modified' and 'Next Steps' are always populated during the summarization lifecycle.
Context Compression Data Schema & Taxonomy
This skill organizes context using a structured Markdown schema to ensure maximum LLM readability and minimal information loss.
| Section | Purpose | Example Data |
|---|---|---|
| Session Intent | High-level goal preservation | "Refactor auth logic" |
| Files Modified | Explicit artifact tracking | "auth.ts (Modified)" |
| Decisions Made | Rationales and constraints | "Used Redis for state" |
| Current State | Testing and status probes | "14/16 tests passing" |
| Next Steps | Task continuity planning | "Fix mock session tests" |
Context Compression Advanced Features
- Anchored Iterative Summarization for non-destructive state merging and persistent context.
- Probe-Based Evaluation to measure functional quality and recall accuracy post-compression.
- Three-Phase Workflow support specifically for Research, Planning, and Implementation phases.
- Specialized artifact index tracking to maintain 100% integrity on file modifications.
- Integration with other Openclaw Skills such as context-degradation and memory-systems for holistic state management.
SKILL.md
Loading
Related Openclaw Skills
ClawGuard
ClawGuard is an LLM-powered security guardrail that evaluates and intercepts risky tool executions to protect your agent environment.
Giraffe Guard
A comprehensive security auditing tool designed to scan Openclaw Skills for malicious code, supply chain vulnerabilities, and sensitive data leaks.
issuefinder-tool
A professional automotive diagnostic tool for Openclaw Skills that automates vehicle log retrieval from the cloud and provides intelligent fault analysis.
Boot KPI Analyzer
An automated diagnostic tool for analyzing the startup and shutdown performance metrics of SS4 vehicle head units.
Error Guard
A proactive error management and circuit-breaking system designed to prevent AI agent failures through real-time monitoring and automated recovery.
Proactive Agent
A self-improving agent architecture designed to anticipate user needs, maintain state across context loss, and operate autonomously.
Featured*
