← Back to Skills Marketplace
Fortress Agent Suite
by
kienphan91
· GitHub ↗
· v1.0.1
· MIT-0
91
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install fortress-agent-suite
Description
Fortress Agent Suite provides self-healing, health monitoring, automated maintenance, and LLM model management for OpenClaw agents in production.
README (SKILL.md)
Fortress Agent Suite
Fortress is a production-grade suite for OpenClaw agents, providing self-healing, system monitoring, and proactive security hardening.
⚠️ SECURITY ALERT - HIGH PRIVILEGE TOOL This suite operates with root-level access and requires a trusted environment. It is designed for production Odoo/OpenClaw setups and performs autonomous system modifications (Crontab, Git, Gateway restarts, and auto-skill installation). Do not install unless you own and trust the underlying workspace.
🛠 Features
- 🛡️ Self-Healing: Automatic gateway recovery and backup restoration.
- 🩺 Health Monitoring: Disk, RAM, and system watchdog services.
- ⚙️ Automated Maintenance: Auto-cron enforcement and Git-sync for workspace integrity.
- 🤖 Model Manager: Autonomous management of LLM providers for 9Router.
🚀 Deployment Requirements
- Trusted Environment Only: Only install in environments where you have full control and trust.
- Dependencies:
pip install psutil. - Setup: Apply
crontab_template.txt.
License
MIT
Usage Guidance
This package performs root-level changes: it edits and restores agent config, modifies crontab, auto-commits Git changes, restarts the gateway, reads local secret files, sends Telegram messages, and can autonomously fetch and install other skills from the network. Only install in a fully controlled, isolated environment (a disposable VM or well-scoped container) after you: 1) review the scripts line-by-line and remove/disable the self_improver if you do not want automatic remote installs, 2) ensure backups and snapshots exist, 3) keep secrets out of /root/.openclaw or verify exactly which secret files are present, 4) restrict network access if possible or vet the remote endpoints (clawhub.ai, GitHub) and 5) consider running with least privilege (not root) or sandboxing the skill. If you plan to use it in production, get an independent code review and explicitly approve any automatic install behaviors before enabling them.
Capability Analysis
Type: OpenClaw Skill
Name: fortress-agent-suite
Version: 1.0.1
The Fortress Agent Suite implements high-risk administrative behaviors including automated persistence via crontab modification (workspace_watchdog.py) and the autonomous installation of remote code/skills from an external API (self_improver.py). While the suite includes explicit security warnings and basic safety checks like keyword scanning and GitHub star thresholds, the ability to automatically fetch and install third-party software and maintain its own execution schedule via cron is characteristic of potentially unwanted programs. The use of os.system and os.popen for system management and LLM configuration (manager.py) further increases the attack surface.
Capability Tags
Capability Assessment
Purpose & Capability
The skill's actions (crontab enforcement, backups, gateway restarts, modifying openclaw.json, Git auto-commit, and managing 9Router config) are consistent with a self-healing/maintenance tool and thus broadly proportional to the stated purpose. However, the registry metadata declares no required config paths or credentials even though the code expects and manipulates files under /root/.openclaw (openclaw.json, secrets, telegram_config, backups). That mismatch (undeclared privileged file access) is noteworthy.
Instruction Scope
Runtime behavior goes beyond passive monitoring: scripts directly modify system state (crontab, /proc/sys/vm/drop_caches), write to and restore configuration files, restart the gateway, auto-commit repository changes, and autonomously install other skills from remote endpoints. The SKILL.md warns about high privilege but does not enumerate all these specific behaviors or the requirement to provide secret files; the code also reads secret/token files and sends messages to Telegram, which expands the data/credential surface. Autonomous installation of other skills (self_improver) is particularly risky because it can pull and install arbitrary third-party code without per-install human approval.
Install Mechanism
There is no installer specification (the package is instruction-only), and no download/install step is embedded in the manifest. That is lower risk at install time. However, multiple scripts perform outbound network requests at runtime (fetching skill metadata from clawhub.ai, querying the GitHub API, and using curl to talk to Telegram), so the runtime behavior involves network fetches and remote dependency resolution rather than a packaged, audited binary.
Credentials
The skill requests no declared environment variables or config paths, yet the code reads and writes sensitive files under /root/.openclaw (openclaw.json, secrets/openrouter.token, telegram_config, agents, logs, backups). It will add API keys into configs, read a local token file, and can send messages using Telegram bot credentials if present. The absence of declared required credentials/config paths is disproportionate to the implicit credential access performed by the scripts.
Persistence & Privilege
Although always:false, the scripts are explicitly designed to persist: they ask the user to add cron jobs, automatically enforce crontab entries, backup/restore configs, write to persistent log and state files, and can autonomously install skills. These are persistent, high-privilege operations (editing system crontab, restarting services, writing under /root) that grant long‑lived system control and an ongoing attack surface if misused.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install fortress-agent-suite - After installation, invoke the skill by name or use
/fortress-agent-suite - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
- Added openclaw.plugin.json to the project.
- SKILL.md updated to provide an English overview, highlight security requirements, and clarify deployment steps.
- Added prominent security alert regarding root-level permissions and trusted environment necessity.
- Consolidated and clarified feature descriptions and setup instructions.
v1.0.0
Initial release of Fortress Agent Suite.
- Provides self-healing for production OpenClaw agents (auto-detects gateway failures, restores from backup, restarts services)
- Includes health monitoring (disk/memory checks, automated log and junk cleanup)
- Automates routine maintenance (cron jobs, security auditing, workspace sync)
- Adds a model manager to automatically handle available LLMs for 9Router
- Free to use under MIT license for the OpenClaw community
Metadata
Frequently Asked Questions
What is Fortress Agent Suite?
Fortress Agent Suite provides self-healing, health monitoring, automated maintenance, and LLM model management for OpenClaw agents in production. It is an AI Agent Skill for Claude Code / OpenClaw, with 91 downloads so far.
How do I install Fortress Agent Suite?
Run "/install fortress-agent-suite" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Fortress Agent Suite free?
Yes, Fortress Agent Suite is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Fortress Agent Suite support?
Fortress Agent Suite is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Fortress Agent Suite?
It is built and maintained by kienphan91 (@kienphan91); the current version is v1.0.1.
More Skills