A proactive local skill manager that detects, extracts, and maintains reusable AI agent behaviors from your development sessions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install autoskill-local-skill-manager
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 autoskill-local-skill-manager 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 autoskill?
autoskill is a sophisticated management layer for Openclaw Skills designed to turn transient conversation insights into durable, reusable agent capabilities. It acts as a lightweight self-improving memory system that monitors your sessions for unique workflows, project-specific conventions, and personal preferences, ensuring that valuable automation patterns are never lost.
Unlike generic repositories, this skill focuses on preserving the technical truth of your workspace. It operates directly on your local filesystem, organizing Openclaw Skills into structured folders containing markdown documentation, agent metadata, and supporting scripts. By automating the lifecycle of skill discovery and refinement, it allows your AI agents to evolve alongside your specific technical requirements without manual overhead.
autoskill Use Cases
- Automatically detecting reusable code patterns or workflow corrections during an active coding session.
- Merging new project-specific style guides or naming conventions into existing Openclaw Skills.
- Identifying when a sequence of manual terminal commands should be converted into a reusable script-based skill.
- Deduplicating similar automation attempts by searching through local and external skill ecosystems.
- Refining agent trigger descriptions to prevent over-triggering or under-triggering in complex environments.
How autoskill Works
- Silent Scanning: After every substantive interaction, the system asks if the turn contains reusable personal preferences, rules, or workflows.
- Extraction Check: When a trigger is identified, the system classifies the experience and searches for existing Openclaw Skills to avoid duplication.
- Candidate Drafting: A proposed skill or diff is drafted, including audience classification (personal, team, or public) and priority scoring.
- User Confirmation: All proposed changes, including file paths and unified diffs, are presented to the user for explicit approval.
- Execution: Upon approval, the system creates or updates the local skill directory structure, including SKILL.md and optional assets or scripts.
autoskill Setup
Identify your local skill root and ensure the directory is accessible to your agent. You can initialize new Openclaw Skills using the following structure:
# Example of creating a skill directory manually
mkdir -p ~/.codex/skills/my-new-skill
touch ~/.codex/skills/my-new-skill/SKILL.md
# If skill-creator tools are available, autoskill can use them:
python3 <skill-creator>/scripts/init_skill.py my-new-skill --path ~/.codex/skills
autoskill Data Schema & Taxonomy
The skill organizes data into standard Openclaw Skills folders. Each folder follows a strict taxonomy to ensure portability and clarity:
| File/Folder | Purpose |
|---|---|
SKILL.md |
The core definition, including YAML frontmatter and imperative instructions. |
agents/openai.yaml |
(Optional) Model-specific configuration and metadata. |
scripts/ |
(Optional) Reusable, deterministic automation scripts. |
references/ |
(Optional) Bulky documentation, checklists, or templates. |
assets/ |
(Optional) Static files used as output components. |
autoskill Advanced Features
- Multi-layer trigger policy including silent scans, extraction checks, and scheduled session-end reviews.
- Sophisticated de-duplication logic that compares skills by task family, tool usage, and output contracts.
- Language consistency engine that matches the skill's primary language to the user's input evidence.
- Background execution mode that drafts improvements without blocking the primary user task.
- Comprehensive validation loops that use realistic prompts to test trigger accuracy and instruction clarity.
SKILL.md
Loading
Related Openclaw Skills
Rental Car Return Checklist
A comprehensive documentation and inspection framework designed to protect renters from unfair cleaning, damage, or fuel fees upon vehicle return.
D E Shaw Research Skill
A comprehensive intelligence skill for researching the history, business models, and quantitative strategies of D E Shaw Research.
Cognizant Intelligence & IT Strategy
A specialized research skill for analyzing the business models, market positioning, and operational scale of Cognizant within the IT services landscape.
Dannon Market Analysis
A specialized technical skill for analyzing the North American yogurt market evolution and Danone's global branding strategies.
Circle K Stores Analysis Skill
A comprehensive research skill for analyzing the global operations, acquisition history, and business model of Circle K and its parent company, Alimentation Couche-Tard.
Flight Delay Compensation Prep
An automated assistant designed to analyze flight disruptions and generate structured compensation claims based on global aviation regulations.
Featured*
