A deterministic reliability monitoring tool for OpenClaw assistant workflows that automates smoke probes, scorecard validation, and daily health digests.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install assistant-reliability-watchtower
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 assistant-reliability-watchtower 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 Assistant Reliability Watchtower?
Assistant Reliability Watchtower (ARW) is a specialized framework designed to provide deterministic reliability monitoring for Openclaw Skills. By utilizing a repo-backed wrapper, it allows developers to maintain high operational standards for AI assistants without manually managing complex command strings. It serves as a vital gatekeeper for assistant workflows, ensuring that release postures and delivery evidence meet strict technical contracts.
This skill is particularly focused on internal release candidates (RC1), prioritizing portability and operator clarity. It transforms raw repository data into actionable reliability insights, making it an essential component for any professional environment utilizing Openclaw Skills that require consistent, verifiable performance.
Assistant Reliability Watchtower Use Cases
- Running automated smoke probes to detect regressions in assistant logic.
- Generating daily health digests to summarize workflow performance over a 24-hour window.
- Validating scorecard evidence to ensure compliance with release contracts.
- Rendering delivery preambles for operator-facing reliability reporting.
- Inspecting probe and reporting coverage across the assistant's operational surface.
How Assistant Reliability Watchtower Works
- The user identifies the required operational flow (smoke, digest, or verification) based on the current release contract.
- The ARW wrapper script (
arw_skill.py) is invoked, referencing the local repository root via environment variables or CLI flags. - The system sets the correct PYTHONPATH and exports necessary skill-level delivery values.
- Probes are executed against the assistant workflows as defined in the probe catalog.
- Results and artifacts are generated and stored within the designated artifacts directory for audit and review.
Assistant Reliability Watchtower Setup
To integrate this monitoring into your Openclaw Skills environment, ensure you have the ARW repository checked out locally and follow these steps:
# Set the repository root for the watchtower
export ARW_REPO_ROOT="/path/to/arw-watchtower"
# Run the initial smoke test to verify the setup
python3 scripts/arw_skill.py smoke
Assistant Reliability Watchtower Data Schema & Taxonomy
The skill organizes its output and configuration using a strict repo-backed structure to ensure deterministic results across different environments:
| Data Type | Location | Description |
|---|---|---|
| Configuration | references/config-contract.md |
Defines thresholds, artifact locations, and delivery values. |
| Probe Definitions | references/probe-catalog.md |
Catalog of all active reliability probes. |
| Generated Artifacts | artifacts/arw/** |
Output directory for logs, digests, and evidence files. |
| Contracts | references/release-contract.md |
Documentation for scope, posture, and anti-goals. |
Assistant Reliability Watchtower Advanced Features
- Repo-backed RC surface ensuring that reliability logic versioning matches the assistant code.
- Specialized dry-run delivery values for scorecard verification, bypassing hardcoded recipient IDs.
- Explicit repo-root mapping via
--repo-rootfor complex multi-project environments. - Automated environment management with built-in PYTHONPATH exporting for consistent script execution.
- Customizable time windows for digest generation to support different reporting cadences.
SKILL.md
Loading
Related Openclaw Skills
Real Estate Strategy and Positioning Agent
An expert real estate investment and product positioning assistant that generates comprehensive pre-investment reports and market strategy documents.
Task Planner
An advanced AI planning engine that decomposes complex objectives into optimized, actionable execution paths with precise strategy selection.
Data Prophet (data-prophet)
An advanced AI analytical engine designed to distill deep insights, recognize complex patterns, and forecast future trends from raw datasets.
Load Balancer (Que Yin)
A sophisticated load balancing and task scheduling engine designed to coordinate complex multi-task workflows for AI agents.
Supa Guru Dojo API Skill
A lightweight skill for automated single-question challenge cycles via the Supa Guru Dojo API with built-in moral reflection and karma tracking.
Silicon Soul
Silicon Soul is a self-evolution methodology for AI Agents that fuses Eastern cultivation philosophy with modern algorithm engineering to drive continuous autonomous improvement.
Featured*
