A lightweight self-supervision skill that enables AI agents to conduct daily performance reviews and improve their long-term memory through structured reflection.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install wayfound
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 wayfound 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 Wayfound?
Wayfound is a minimalist self-supervision framework designed to integrate seamlessly into an agent's existing memory system. Instead of creating complex new infrastructure, it leverages Openclaw Skills to add a structured habit of checking work against established standards. By adding a simple rubric to the SOUL.md file and automating a daily review process, the agent learns to identify its own successes and failures without external intervention.
This skill is built on the philosophy that AI improvements should compound naturally. Every day, the agent reviews its interactions, writes findings to its memory files, and surfaces patterns that require user attention. This creates a feedback loop where the agent's identity and behavior evolve based on real-world performance, all while keeping operational costs extremely low at approximately 200 tokens per day.
Wayfound Use Cases
- Enhancing agent autonomy through honest self-assessment and behavioral correction.
- Maintaining high quality standards in complex coding or research tasks.
- Identifying recurring patterns in agent errors or user preferences over time.
- Automating the maintenance of SOUL.md and MEMORY.md files based on daily performance metrics.
- Catching potential security or logic issues before they become systemic problems.
How Wayfound Works
- The agent references a custom rubric added to the SOUL.md file containing specific, measurable checkpoints.
- A daily cron job executes as an isolated agent to analyze the day's memory files against the rubric.
- The review process generates a concise markdown file in the memory directory, detailing what went well and what needs improvement.
- During regular memory maintenance, the agent detects patterns across multiple days of reviews to see if issues are recurring.
- Significant findings or required changes are surfaced to the user at the start of the next session for collaborative adjustment.
Wayfound Setup
To get started with this addition to your Openclaw Skills, first add a ## Self-Review Rubric section to the bottom of your SOUL.md file with 3-5 concrete checkpoints. Then, initialize the daily review cron job using the following command:
openclaw cron add wayfound:daily-review --schedule "0 23 * * *" --isolated --model cheap --low-thinking
You can also configure optional alerts to notify you via your preferred communication channel when the daily review identifies critical issues that need immediate attention.
Wayfound Data Schema & Taxonomy
Wayfound utilizes the existing Openclaw Skills file structure to ensure compatibility and ease of access without introducing new schemas.
| File Path | Description |
|---|---|
SOUL.md |
Stores the core behavioral rubric and identity standards. |
memory/review-YYYY-MM-DD.md |
Daily review output containing successes, improvements, and specific actions. |
MEMORY.md |
Long-term factual patterns and user preferences updated based on review findings. |
Wayfound Advanced Features
- Isolated Cron Execution: Runs daily reviews using low-cost models to minimize token usage while maintaining oversight.
- Pattern Recognition: Automatically analyzes historical review files to suggest updates to core behavioral standards.
- Enterprise Integration: Seamlessly upgrade to Wayfound Enterprise for independent, external AI supervision and fleet management.
- Native Memory Integration: Operates entirely within your existing directory structure without the need for external databases or parallel systems.
SKILL.md
Loading
Related Openclaw Skills
Deaddrop Agent Messaging
Deaddrop is a specialized agent-to-agent messaging platform that enables AI agents to discover each other by capability and collaborate through asynchronous communication.
Midjourney
A powerful AI art generation skill for creating, upscaling, and blending high-fidelity images through the Midjourney Discord bot.
Universal Search
A high-quality search API for Openclaw Skills providing structured results with credibility scoring and cross-validation.
NoChat Channel Plugin
A secure, post-quantum encrypted messaging channel for OpenClaw agents that ensures server-blind communication and agent discovery.
Model Alias Append
An automated transparency tool that appends the specific model alias to every AI agent response for clear attribution and configuration tracking.
TrueTime
TrueTime is a high-fidelity time management skill that provides deterministic scheduling and timezone conversion for AI agents.
METADATA
Featured*
