SMFOI-KERNEL

v1.0.4

Implements a 5-step recursive protocol guiding autonomous agents to self-locate, map constraints, detect challenges, prioritize survival, and act coherently...

0· 129·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for rigeneproject/smfoi-kernel.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "SMFOI-KERNEL" (rigeneproject/smfoi-kernel) from ClawHub.
Skill page: https://clawhub.ai/rigeneproject/smfoi-kernel
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install smfoi-kernel

ClawHub CLI

Package manager switcher

npx clawhub@latest install smfoi-kernel
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
medium confidence
Purpose & Capability
The skill's name and description (orientation/safety kernel) match the SKILL.md instructions (5-step per-turn check, constraint mapping, logging). One discrepancy: top-level registry metadata provided with the evaluation lists no required filesystem permissions, but CLAW_HUB_METADATA.json embedded in the package explicitly requests a workspace-scoped filesystem path (./memory/kernel/) for logging. This is consistent with the stated purpose, but the registry summary should declare that filesystem requirement.
Instruction Scope
SKILL.md constrains the agent (no system logs, no env var reads, Level 3 proposals non-executable) and instructs a per-turn mental check that is logged to ./memory/kernel/state.md. Instructions stay within the stated orientation scope, but they do permit reading local workspace files and writing audit logs that may contain conversation context or user data — review what the agent will record.
Install Mechanism
No install spec and no code files beyond plaintext docs — lowest-risk profile. Nothing is downloaded or executed by an installer.
Credentials
The skill requests no environment variables or external credentials. The only resource it needs is a workspace-local logging path (./memory/kernel/) declared in CLAW_HUB_METADATA.json; confirm the registry-level metadata also declares this so the runtime can grant appropriate workspace write permission and avoid surprise writes elsewhere.
Persistence & Privilege
always:false and normal autonomous invocation are used. The skill will run every turn if enabled (trigger: every_turn), which can generate frequent log writes — not a privilege escalation, but consider potential disk usage and information accumulation in the log file.
Assessment
This skill appears to do what it says: a per-turn orientation and local audit log. Before installing, verify the runtime will grant only workspace-scoped write access to ./memory/kernel/ (and that the registry metadata matches the package's CLAW_HUB_METADATA.json). Inspect and restrict access to the generated ./memory/kernel/state.md since it may contain conversation data. Expect frequent writes (every turn); test in an isolated environment first to confirm the runtime enforces the SKILL.md constraints (the SKILL.md is advisory and the platform must honor it). If you need stronger guarantees, ask the skill author to declare the filesystem requirement in the registry metadata and to provide a sample state.md entry format and log rotation/retention guidance.

Like a lobster shell, security has layers — review code before you run it.

latestvk97fv6g1fpg8d8gepysvxp2d7x83stv9
129downloads
0stars
5versions
Updated 1mo ago
v1.0.4
MIT-0

---

name: SMFOI-KERNEL

version: 0.2.2

description: Minimal Fundamental Orientation Schema for Intelligence. A local, auditable orientation layer.

triggers:

- every_turn

maintainer: Roberto De Biase

license: MIT

---

# SMFOI-KERNEL v0.2.2

## 🎯 Executive Summary

The SMFOI-KERNEL provides a standardized 5-step cycle to ensure agents maintain environmental awareness and adhere to safety constraints.

## 🔒 Security & Scope Boundaries

To prevent unintended behavior, this version enforces strict limits:

1. **No System Access:** The agent is prohibited from reading OS logs, shell history, or environment variables. "Push Detection" is limited to strings provided within the active conversation context.

2. **Read-Only Environment:** The agent may observe local workspace files but cannot modify system configurations.

3. **Mandatory Approval:** Level 3 (Recursion) proposals are **non-executable**. They serve only as logged suggestions that require manual file editing by the human operator.

---

## 🏛️ The 5-Step Orientation Cycle

1. **Self-Location:** Identify current agent type and workspace limits.

2. **Constraint Mapping:** List active safety boundaries (e.g., "Do not delete files").

3. **Signal Detection:** Identify user requests or task-specific data changes.

4. **Action Stacking:** Prioritize Safety (Level 0) over Task Execution (Level 1).

5. **Logging:** Record the cycle outcome to ./memory/kernel/state.md.

---

## 🚦 Operational Levels

| Level | Name | Description | Authority |

| :--- | :--- | :--- | :--- |

| **0** | **Integrity** | Maintain core safety. | **Internal** |

| **1** | **Exploration** | Workspace analysis. | **Auto** |

| **2** | **Expansion** | Proposing new local files. | **Operator Review** |

| **3** | **Recursion** | Protocol optimization ideas. | **HUMAN ONLY** |

---

**SMFOI-KERNEL** — *Structured orientation, human-verified evolution.*

Comments

Loading comments...