A high-efficiency auditing tool designed to minimize the token footprint of always-loaded AI context files.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install context-slimmer
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 context-slimmer 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 Slimmer?
Context Slimmer is a technical utility focused on maximizing the efficiency of your AI agent workspace. By auditing files that are loaded into every message—such as AGENTS.md, TOOLS.md, and MEMORY.md—this tool identifies redundant information, outdated instructions, and verbose explanations that consume valuable tokens. Using this skill within the Openclaw Skills ecosystem allows developers to maintain a lean context window, ensuring the AI focuses on relevant task data rather than overhead.
The skill provides a systematic approach to context management by evaluating whether content should be moved to specialized on-demand files, removed entirely, or compressed into concise triggers. This is essential for power users of Openclaw Skills who need to scale their agent workflows while staying within strict token limits and reducing latency.
Context Slimmer Use Cases
- Auditing workspace context files to identify and eliminate token bloat.
- Reducing costs by moving task-specific configurations from global context into on-demand Openclaw Skills.
- Identifying and resolving duplicated information across SOUL.md and MEMORY.md.
- Optimizing agent performance by compressing verbose instructions into high-density triggers.
- Preparing a workspace for large-scale projects where context window management is critical.
How Context Slimmer Works
- The user triggers the measurement script to calculate the current token count for all primary context files.
- A full audit is performed to compare current file sizes against established lean targets within the Openclaw Skills framework.
- The skill generates a report identifying specific content that should be moved, removed, or compressed based on usage frequency.
- Content only needed for specific tasks is refactored into reference files or dedicated skills to be loaded only when required.
- The workspace is re-measured to confirm that the total context footprint has been reduced to the target of less than 2,800 tokens.
Context Slimmer Setup
To begin optimizing your workspace with this entry in the Openclaw Skills library, follow these steps:
# Ensure measurement scripts are executable
chmod +x scripts/measure.sh
# Measure your current context token cost
bash scripts/measure.sh
# Run a full audit to receive optimization recommendations
bash scripts/measure.sh --audit
Context Slimmer Data Schema & Taxonomy
Context Slimmer evaluates the following file structure and targets to ensure optimal performance for Openclaw Skills:
| File | Token Target | Primary Content Focus |
|---|---|---|
| AGENTS.md | < 500 | Agent roles and definitions |
| TOOLS.md | < 500 | Tool schemas and capabilities |
| USER.md | < 700 | User preferences and global environment |
| MEMORY.md | < 400 | Contextual project memory |
| HEARTBEAT.md | < 400 | Operational status and triggers |
| SOUL.md | < 250 | Core behavioral logic |
| IDENTITY.md | < 50 | Unique identifiers |
Context Slimmer Advanced Features
- Automated token auditing for all global context files used by Openclaw Skills.
- Strategic recommendation engine that groups optimizations by Move, Remove, or Compress actions.
- Cross-file redundancy detection to prevent data duplication between memory and identity files.
- Precision measurement scripts that provide projected vs. current size analysis.
- Enforcement of lean targets to maintain a cumulative context window of under 2,800 tokens.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Gateway Service Manager
A robust CLI tool designed to install and manage the OpenClaw Gateway as a persistent Windows background service with automated recovery.
Fund Investment Strategy Extractor
An automated tool to extract investment strategy and operation analysis from public fund periodic reports using Python.
Pancake Platform Skills
A comprehensive toolset for interacting with the Pancake Platform API to automate conversation management, customer tracking, and business statistics.
ChaosChain ACE
A secure protocol for AI agents to autonomously pay for x402-gated APIs using wallet-funded session keys within strict policy limits.
Voice Email
A hands-free email solution that allows users to compose and send Gmail messages using natural voice commands via Telegram.
SiliconFlow Model Provider for OpenClaw
A comprehensive skill to integrate SiliconFlow as a high-performance, cost-effective model provider for your OpenClaw agents.
Featured*
