A security-focused execution layer that makes risky AI agent behaviors observable, verifiable, and recoverable through a hardened runtime.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install provenable
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 provenable 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 AEGX Provenable: Hardened AI Agent Runtime?
AEGX Provenable is a comprehensive, hardened runtime environment designed specifically for agentic systems. It transforms static safety policies into active, verifiable guardrails. By providing a dedicated control surface for Openclaw Skills, it ensures that high-risk actions—such as file system changes, network egress, and skill installations—are performed under constant supervision. The system operates through a dual-layered approach: a foreground CLI for operator control and a background daemon that maintains live protection, heartbeats, and system health visibility.
The core identity of AEGX Provenable is its ability to make risky agent behavior observable rather than relying on static policy text alone. It provides a robust architecture where the skill acts as the front door, while the runtime and daemon serve as the always-on machine room, ensuring every action taken by an agent is verifiable and, if necessary, recoverable.
AEGX Provenable: Hardened AI Agent Runtime Use Cases
- Integrating hardened runtime guards into Openclaw Skills to prevent unauthorized system mutations.
- Ensuring safe operation by verifying daemon health and system readiness before executing trust-sensitive tasks.
- Conducting forensic incident investigations using tamper-evident evidence bundles and audit-grade reports.
- Managing system state through snapshots and rollbacks to recover from unsafe agent behaviors or policy violations.
- Protecting sensitive data planes including memory, conversation context, and network egress points.
How AEGX Provenable: Hardened AI Agent Runtime Works
- The operator builds the Rust workspace and initializes the state and policy directory using the command-line interface.
- The background daemon is started to provide continuous monitoring, heartbeat tracking, and readiness visibility.
- Before any risky action, the system performs an active self-verification check to confirm all protection layers are live.
- The agent executes tasks through guarded paths where the guard engine evaluates actions against control, memory, and network surfaces.
- Snapshots are created automatically or manually before destructive changes to provide a secure rollback path.
- In the event of a policy violation or incident, the system exports a signed evidence bundle for post-incident review and integrity verification.
AEGX Provenable: Hardened AI Agent Runtime Setup
To install and configure the system for Openclaw Skills, follow these steps:
# Build the Rust workspace
cargo build --workspace --release
# Initialize state, policy, and workspace directories
aegx init
# Verify base system health and daemon connectivity
aegx status
aegx daemon status
# Run active self-verification before starting agent tasks
aegx self-verify --active
AEGX Provenable: Hardened AI Agent Runtime Data Schema & Taxonomy
The skill organizes data into a structured hierarchy to maintain a verifiable audit trail and system state.
| Component | Data Type | Description |
|---|---|---|
| Evidence Store | Binary / JSON | Persistent records and audit-chain mechanics for post-incident review. |
| Snapshots | State Image | Captured system states used for rollback and recovery logic. |
| Taint Labels | Metadata | Labels used to track the propagation of untrusted data across surfaces. |
| Audit Chain | Hash-linked List | A tamper-evident chain of records documenting all principal actions. |
| Guard Policies | JSON Schema | Machine-readable definitions for file, network, and output protection planes. |
AEGX Provenable: Hardened AI Agent Runtime Advanced Features
- Multi-platform adapter support specifically optimized for Openclaw Skills, Manus, and Claude Code.
- Granular guard surfaces covering control, memory, conversation, file-read, and network egress planes.
- Automated degraded mode detection that halts high-risk actions if background protection becomes stale or unreachable.
- Tamper-evident evidence bundling with built-in integrity verification and human-readable reporting tools.
- Advanced rollback policy engine that restores protected state while maintaining an audit trail of the recovery.
SKILL.md
Loading
Related Openclaw Skills
Phase Theorist
A specialized AI research agent grounded in the complete Phase Theory repository for precise interpretation, technical derivation, and ontological reasoning.
OpenClaw Multi-Agent Wizard
A guided, low-jargon installation wizard that simplifies the setup of multiple AI agents and Feishu bot bindings.
Knowledge Lib
A sophisticated personal knowledge wiki manager that transforms raw links and documents into a structured, cross-referenced Markdown knowledge base.
Nex Deliverables
A local-first client deliverable tracking and project management system designed for agencies and freelancers to monitor project lifecycles and deadlines.
Feishu All-in-One Pro
A comprehensive automation suite that unifies all Feishu office functions into a single, high-efficiency interface for AI agents.
Finance Management Skill
A comprehensive AI-powered financial management skill for tracking expenses, income, and bank statements through natural language and file processing.
Featured*
