A lightweight ground-truth layer that injects verified facts directly into an AI agent's system context to eliminate hallucinations.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-plugin-edicts
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 openclaw-plugin-edicts 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 Edicts: Ground Truth for AI Agents?
Edicts is a specialized plugin designed to provide a reliable memory layer for AI agents. By integrating with Openclaw Skills, it allows developers to maintain a set of immutable or verified facts—such as product launch dates, compliance rules, and naming conventions—that are automatically injected into every prompt. This ensures that the agent operates within the bounds of reality without the need for a complex retrieval-augmented generation (RAG) stack.
Built with efficiency in mind, Edicts supports both YAML and JSON formats and includes a configurable token budget to ensure that ground-truth data doesn't overwhelm the agent's context window. It serves as a guardrail, ensuring that your Openclaw Skills driven agents remain accurate, professional, and compliant with organization-specific requirements.
Edicts: Ground Truth for AI Agents Use Cases
- Enforcing strict product launch dates and embargoes to prevent premature disclosures.
- Maintaining consistent naming conventions and branding rules across generated content.
- Injecting real-time compliance rules and security protocols into agent workflows.
- Tracking migration status and system updates to keep technical agents informed.
- Managing engagement rules for social media or community interaction agents.
How Edicts: Ground Truth for AI Agents Works
- Definition: You define verified facts (edicts) in a local configuration file with metadata like confidence levels and TTL.
- Injection: The plugin automatically reads these edicts and formats them into the system context of every prompt sent to the AI.
- Discovery: The agent uses built-in tools to search or list available edicts to find specific information relevant to the current task.
- Mutation (Optional): If enabled, the agent can update or add new edicts during a session to reflect changing project states.
- Cleanup: Stale or expired edicts are automatically identified for removal during review cycles to keep the context clean.
Edicts: Ground Truth for AI Agents Setup
Install the plugin via the CLI using Openclaw Skills management commands:
openclaw plugins install openclaw-plugin-edicts
Then, configure the plugin in your openclaw.json file:
{
"openclaw-plugin-edicts": {
"enabled": true,
"config": {
"format": "yaml",
"maxEdicts": 200,
"tokenBudget": 4000,
"includeSystemContext": true
}
}
}
Edicts: Ground Truth for AI Agents Data Schema & Taxonomy
The skill organizes data using a structured format that includes metadata for better context management:
| Property | Description | Possible Values |
|---|---|---|
text |
The core fact or rule | String |
category |
Classification of the edict | product, compliance, dev, etc. |
confidence |
The reliability of the fact | verified, inferred, user |
ttl |
Retention policy | ephemeral, event, durable, permanent |
expiresAt |
Expiration timestamp | ISO Date String |
tags |
Searchable keywords | Array of Strings |
Edicts: Ground Truth for AI Agents Advanced Features
- Token Budgeting: Automatically prunes context injection to stay within a defined token limit (default 4000).
- Opt-in Write Access: Secure by default with read-only tools; write operations require explicit configuration.
- Stale Thresholds: Configurable days-to-stale settings to identify and clean up outdated information.
- Multi-Format Support: Choice between YAML for human readability or JSON for machine interoperability.
- System Context Injection: Toggleable prompt injection for cases where only tool-based lookup is desired.
SKILL.md
Loading
Related Openclaw Skills
Humanizer
The humanizer skill refines AI-generated text by identifying and removing common algorithmic patterns to create more natural, authentic writing.
Knitting Pattern Search Skill
A specialized AI agent skill designed to programmatically find and retrieve knitting patterns for craft-focused workflows.
Stock Tools
A streamlined A-share market data tool for managing personal watchlists and fetching real-time price quotes through lightweight public APIs.
PDF to LaTeX
A specialized tool that leverages the MinerU engine to transform complex PDF documents, including academic papers and technical reports, into editable LaTeX source code.
Nano Banana Pro
A high-performance image generation and editing skill powered by Gemini 3 Pro Image via OpenRouter.
Google Postmaster Tools for AI Agents
A powerful integration for AI agents to monitor Gmail sender reputation and email deliverability metrics through the Membrane ecosystem.
Featured*
