Sigil Security provides a robust 3-layer Guardian validation and co-signing framework for securing ERC-4337 smart wallets used by AI agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install sigil-security
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 sigil-security 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 Sigil Security?
Sigil Security is a specialized security skill designed to protect autonomous agent wallets through the Sigil Protocol. By leveraging this tool within your Openclaw Skills ecosystem, developers can deploy secure ERC-4337 smart accounts across six major EVM chains, including Ethereum, Polygon, and Base. The core value lies in its 3-layer Guardian architecture, which subjects every agent transaction to a rigorous validation process—comprising rule-based checks, transaction simulation, and AI-driven risk scoring—before any funds are moved.
This skill separates administrative control from operational execution. While the human owner maintains full authority over policies and whitelists, the AI agent is granted a limited-capability signer. This ensures that even if an agent's environment is compromised, the Guardian prevents unauthorized transactions that violate the owner's pre-set security boundaries. Integrating these Openclaw Skills into your workflow provides the necessary safety rails for high-stakes autonomous financial operations.
Sigil Security Use Cases
- Protecting DeFi assets managed by autonomous agents with daily spending limits.
- Restricting AI agents to interact only with specific whitelisted smart contracts and function selectors.
- Implementing multi-chain wallet management for agents across Polygon, Arbitrum, Base, and Avalanche.
- Monitoring agent transaction history and risk scores for auditing and compliance purposes.
- Rotating agent signing credentials instantly without needing to migrate funds.
How Sigil Security Works
- The AI agent generates a User Operation (UserOp) and signs it locally using a dedicated agent signer EOA.
- The signed UserOp is transmitted to the Sigil API endpoint for validation.
- The Guardian service evaluates the transaction against the owner's security policy, including value limits and whitelists.
- A simulation is performed to ensure the transaction will not fail or cause unexpected state changes.
- An AI risk score is generated to detect anomalies or suspicious patterns in the transaction data.
- If all checks pass, the Guardian co-signs the transaction and submits it to the blockchain via the ERC-4337 EntryPoint.
Sigil Security Setup
To integrate this into your Openclaw Skills setup, follow these steps:
- Generate an API key at the Sigil dashboard (starts with
sgil_). - Deploy a Sigil smart account via the dashboard and copy its address.
- Create a dedicated Agent Signer (EOA) and fund it with a minimal amount of gas for the target chain.
- Set the following environment variables:
export SIGIL_API_KEY="sgil_your_key_here"
export SIGIL_ACCOUNT_ADDRESS="0xYourSigilAccountAddress"
export SIGIL_AGENT_SIGNER="0xYourAgentSignerPrivateKey"
- Ensure your local configuration file has restricted permissions using
chmod 600to protect these credentials.
Sigil Security Data Schema & Taxonomy
The skill manages data across several key dimensions to ensure transparency and security:
| Data Type | Description |
|---|---|
| Account Metadata | Stores the Sigil account address, owner address, and associated chain ID. |
| Policy Rules | Includes target whitelists, function selector whitelists, and transaction value limits. |
| Transaction Logs | Detailed history of submitted UserOps, their verdicts, and on-chain transaction hashes. |
| Guardian Reports | Metadata from the 3-layer validation, including AI risk scores and simulation results. |
| API Scopes | Defines granular permissions for the agent signer, such as tx:submit and wallet:read. |
Sigil Security Advanced Features
- Dual-signature enforcement requiring both agent and Guardian signatures for every transaction.
- Zero-admin privileges for the agent signer, preventing unauthorized policy changes or account freezes.
- Support for six EVM-compatible chains including 0G and Avalanche.
- Dry-run evaluation endpoint to test transaction validity without spending gas.
- Automated session key management for routine operations with expiration dates.
- Instant credential rotation through the dashboard to mitigate suspected compromises in Openclaw Skills deployments.
SKILL.md
Loading
Related Openclaw Skills
dispatchi
A robust runner for non-blocking, interactive Claude Code tasks utilizing slash commands and completion callbacks.
Cancel Dispatch Run
A vital utility for stopping active AI dispatch tasks and ralph-loop sessions through a simple slash command.
Explain Code
A visualization and teaching tool that breaks down complex code into relatable analogies, ASCII diagrams, and step-by-step walkthroughs.
The Prompting Company (TPC) Brand Visibility Skill
The Prompting Company skill provides comprehensive brand analytics and Share of Voice tracking across major AI engines like ChatGPT, Claude, and Gemini.
Sardis Balance
A secure, read-only analytics skill for monitoring Sardis agent wallet balances and transaction history.
Sardis Cards
A comprehensive virtual card issuance and management system designed to let AI agents handle real-world transactions with strict programmatic controls.
METADATA
Featured*
