← Back to Skills Marketplace
AGENT-OS
by
Hlias Staurou
· GitHub ↗
· v1.0.3
· MIT-0
451
Downloads
1
Stars
1
Active Installs
3
Versions
Install in OpenClaw
/install agnt-os
Description
The operating system layer for AI agents. Routes goals to the right skills. Executes with checkpoints.
README (SKILL.md)
AGENT-OS
The operating system layer for AI agents.
Routes your goal to the right skill. Executes with checkpoints. Verifies the result.
AGENT-OS v1.0.3 by @contrario
Usage Guidance
This skill is minimal and high-level: it claims to route goals and manage checkpoints but provides no concrete runtime rules or constraints. Before installing, ask the author for a detailed SKILL.md that specifies: (1) exactly how it chooses/calls other skills and whether it will invoke them autonomously; (2) where and how checkpoints are stored and retained; (3) what data is read, logged, or transmitted; and (4) safeguards for sensitive domains (they already mark some domains as 'not recommended'). If you need strict control over which skills can be invoked or what data can be accessed, treat this skill as risky until those details are clarified and test it in a sandboxed environment.
Capability Analysis
Type: OpenClaw Skill
Name: agnt-os
Version: 1.0.3
The skill bundle consists solely of metadata and instructional markdown (SKILL.md) designed to guide the agent's task orchestration. It contains no executable code, external API calls, or file system interactions, and its instructions are consistent with its stated purpose as a routing and verification layer.
Capability Assessment
Purpose & Capability
Name/description claim an 'operating system layer' that routes goals, executes with checkpoints, and verifies results. As an instruction-only skill this could reasonably be a coordination prompt, so the high-level purpose matches. However, nothing in the SKILL.md explains HOW routing or checkpoints are performed, what platform APIs (if any) it expects, or what privileges it requires — leaving a gap between the claimed capability and the declared requirements.
Instruction Scope
The SKILL.md contains only high-level statements (route goals, execute with checkpoints, verify results) and no concrete runtime steps or limits. This is vague and grants the agent broad discretion; it lacks explicit constraints about what data may be read, which skills may be called, how checkpoints are stored, or how verification occurs. Vague, open-ended instructions are a scope creep risk.
Install Mechanism
Instruction-only skill with no install spec and no code files. Lowest-risk installation surface; nothing is written to disk or downloaded by the skill itself.
Credentials
No environment variables, credentials, or config paths are requested — proportional to the instruction-only nature. There is no direct evidence the skill asks for unrelated secrets.
Persistence & Privilege
The skill is not forced-always and does not request elevated persistence. However, because it claims to be an 'OS layer' for routing and verification, if the platform allows autonomous invocation of skills, this role could have broad impact; the SKILL.md does not document intended invocation frequency, checkpoint retention, or limits, which is a potential concern when combined with the skill's high-level role.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install agnt-os - After installation, invoke the skill by name or use
/agnt-os - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.3
v1.0.3: Explicit consent, zero silent actions
v1.0.1
v1.0.1: Transparent boot (no silent actions), first-run consent gate, clarified no filesystem access by AGENT-OS itself
v1.0.0
v1.0.0: Boot sequence, skill orchestration engine, checkpoint system, verified ecosystem — the layer that runs them all.
Metadata
Frequently Asked Questions
What is AGENT-OS?
The operating system layer for AI agents. Routes goals to the right skills. Executes with checkpoints. It is an AI Agent Skill for Claude Code / OpenClaw, with 451 downloads so far.
How do I install AGENT-OS?
Run "/install agnt-os" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is AGENT-OS free?
Yes, AGENT-OS is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does AGENT-OS support?
AGENT-OS is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created AGENT-OS?
It is built and maintained by Hlias Staurou (@contrario); the current version is v1.0.3.
More Skills