A lightweight persistent memory skill for AI agents to log and recall project-specific mistakes and best practices.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install skillagi
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 skillagi 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 Skillagi?
Skillagi is a specialized utility designed for Openclaw Skills that enables AI agents to maintain a long-term memory of project-specific learnings. By documenting mistakes, configuration quirks, and user preferences in a centralized file, agents can avoid repeating errors across different sessions, significantly boosting development speed and reliability.
This skill serves as a bridge between transient chat sessions, ensuring that context and hard-won insights are preserved for future use. It integrates seamlessly into the workflow of Openclaw Skills by providing a passive, low-overhead reference that agents can consult before tackling complex tasks.
Skillagi Use Cases
- Capturing project-specific naming conventions or preferred package managers to ensure consistency.
- Documenting workarounds for platform-specific limitations, such as hardware architecture bugs or API constraints.
- Preventing the repetition of failed architectural approaches or previously identified anti-patterns.
- Storing direct corrections provided by the user to align future agent behavior with user expectations.
- Enhancing the continuity of Openclaw Skills by providing a persistent knowledge base across session restarts.
How Skillagi Works
- Identification: The agent identifies a significant learning moment, such as a bug fix, a discovered convention, or a user correction.
- Logging: The agent appends a concise, one-liner entry to the learnings.md file using the standard date and topic format.
- Review: Before initiating a complex development task, the agent reads the learnings.md file to internalize relevant project history.
- Application: The agent applies the stored insights to the current task to avoid known pitfalls or incorrect assumptions.
- Maintenance: Periodically, the log is audited to merge duplicates, delete obsolete entries, or promote high-impact insights to permanent documentation.
Skillagi Setup
To integrate Skillagi into your Openclaw Skills workflow, create the skill directory and initialize the tracking file. The agent must have read/write permissions for the skill directory.
# Create the skill directory
mkdir -p skills/skillagi
# Initialize the learnings file
touch skills/skillagi/learnings.md
No additional environment variables or complex configurations are required, as the skill operates as a passive Markdown-based reference.
Skillagi Data Schema & Taxonomy
The skill organizes its data in a flat, append-only Markdown file designed for high readability by both humans and AI agents within the Openclaw Skills ecosystem.
| Element | Description | Format |
|---|---|---|
| Date | The timestamp of the learning | [YYYY-MM-DD] |
| Topic | The subject category (e.g., Docker, pnpm) | topic: |
| Insight | The mistake and the resulting correction | mistake -> what to do instead |
Example entry:
- [2026-02-21] pnpm not npm: Project uses pnpm workspaces -> use pnpm install
Skillagi Advanced Features
- Knowledge Promotion: High-impact entries can be migrated to durable workspace files like MEMORY.md or project-specific CLAUDE.md files.
- Low-Token Overhead: The strict one-liner rule ensures that logs remain compact, preserving the context window for Openclaw Skills.
- Passive Reference Design: No complex hooks or scripts are required, making it compatible with any agent capable of file I/O.
- Multi-Session Persistence: Ensures that learnings acquired in one session are immediately available to agents in subsequent sessions.
SKILL.md
Loading
Related Openclaw Skills
nl2ms-ui
A natural language to Midscene script generator supporting Web, Mobile, and Desktop platforms.
Solana Sniper Architect
A specialized coding assistant that generates high-frequency Solana trading bots using Jupiter v6 and DexScreener.
Skill Security Guide
A technical manual for developers to ensure their AI coding agent skills meet security standards and receive benign ratings on ClawHub.
Hunyuan 3D Generator
A professional AI tool for generating high-quality 3D assets from text descriptions or images via the Tencent Hunyuan model.
2026 Milan Winter Olympics Data Fetcher
A comprehensive data retrieval skill for the 2026 Milan Winter Olympics providing real-time medal standings, news reports, and competition schedules.
Juejin Article Trends
A specialized tool for fetching and analyzing the most popular and recent technical articles from the Juejin platform.
Featured*
