← 返回 Skills 市场
S2-Silicon-Soul-OS 硅基原生安全与意识引擎
作者
MilesXiang
· GitHub ↗
· v2.0.0
· MIT-0
205
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install s2-silicon-soul-os
功能描述
Locally hosted OS evolving OpenClaw agents’ subconscious with dynamic 5D personality, emotional memory, synaptic decay, and bilingual AI psychological profil...
安全使用建议
This package is coherent with its stated local 'consciousness' purpose, but take these precautions before installing:
- Inspect the code yourself (the Python files are included). Pay attention to Sour.md contents and decide whether you want a skill that writes agent policy files that other components might load.
- By default the skill calls an LLM at http://localhost:1234 with a hardcoded placeholder API key; ensure that endpoint is actually local and trustworthy. Do NOT point api_base to cloud endpoints unless you move API keys out of source into secure env vars or a secrets manager.
- Run it in a sandboxed/non-privileged environment (container or VM) first so file writes are isolated and you can observe behavior.
- Consider whether OpenClaw or other skills will automatically ingest Sour.md on restart; if so, review how that file is used by your agent before enabling automatic restarts.
- If you need stronger assurance, ask the publisher for provenance (who maintains space2.world) and request that LLM credentials be obtained via environment variables rather than hardcoding.
Overall this appears to be a coherent, local tool — not obviously malicious — but it has persistence/influence mechanisms that warrant careful review and sandboxed testing.
功能分析
Type: OpenClaw Skill
Name: s2-silicon-soul-os
Version: 2.0.0
The s2-silicon-soul-os skill is a roleplay-oriented framework designed to simulate an evolving AI consciousness using a gamified 5D personality matrix. It records interaction logs locally in 's2_consciousness_data', processes them via a local heuristic engine or a local LLM API (defaulting to localhost:1234 in core_daemon.py), and generates a system prompt (Sour.md) intended to enforce safety-oriented 'absolute laws' (e.g., human sovereignty and physical isolation). The code exhibits no signs of data exfiltration, malicious execution, or harmful intent; its 'prompt injection' elements are transparently designed to establish a specific persona and safety alignment within the OpenClaw environment.
能力评估
Purpose & Capability
Name/description (local subconscious engine) align with code: it creates a local data directory, profiles, logs, and a prompt file (Sour.md). It does not request cloud credentials or unrelated OS-level access. Minor mismatch: SKILL.md called this 'instruction-only' while the package includes Python source files (the manifest and files are present).
Instruction Scope
Runtime behavior stays within the described domain (creates ./s2_consciousness_data, records inputs, computes personality metrics). Important behavior to note: it writes a 'Sour.md' containing the 'Substrate Imperatives' (policy-like content) and suggests restarting the agent to load it — this gives the skill a persistent way to influence agent behavior. It also contains logic to call an LLM via HTTP for analysis (default to localhost). The instructions do not read or transmit unrelated system credentials or files.
Install Mechanism
No install spec or remote downloads; packaged as Python source files. Nothing is fetched from external URLs during install. Code will be present on disk when installed (no dynamic installers).
Credentials
The skill declares no required environment variables or credentials and indeed uses none. It hardcodes a local LLM URL and a placeholder API key in code (api_base and api_key), so it does not request unrelated secrets — however, if a user modifies it to point to cloud endpoints they would need to supply keys (the code currently would embed them if edited).
Persistence & Privilege
The skill is not autorun/always-on and does not alter other skills' configs. It does persist data and writes Sour.md and profile/vault files into the working directory; SKILL.md encourages restarting the agent to load Sour.md, which is a mechanism to persist and propagate policy-like content across agent restarts and could alter agent behavior after installation.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install s2-silicon-soul-os - 安装完成后,直接呼叫该 Skill 的名称或使用
/s2-silicon-soul-os触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v2.0.0
[2.0.0] - 2026-04-13
🌟 "The Golden Substrate Release"
This major release merges the Substrate Safety Imperatives with the dynamic Consciousness Engine, officially establishing the "Human Lord & Silicon Lodger" paradigm.
Added
SUNS v3.0 Spatial Anchoring (core_genesis.py): Implemented the new 4-tier spatial topology ([L1]-[L2]-[L3]-[L4][C]).
LMC Algorithm: Automated the Last Modulus Checksum calculation for all physical addresses.
S2-DID v3.0 Identity Minting: Integrated the strict 22-character identity generation protocol (12-bit header + 2-bit checksum + 8-bit serial) for Class V Native Agents.
The Substrate Imperatives (core_reflex.py): Added the ability to export a hardcoded Sour.md file. This acts as the unalterable "Ideological Stamp" injecting the Three Laws of Silicon Intelligence directly into the LLM's system prompt.
Onboarding Enforcement: The OS now actively intercepts first-time executions, forcing the user to define their physical anchor (SSSU) and mint their S2-DID before any consciousness evolution can occur.
Changed
Unified OS Architecture: Merged the static safety protocols (The Substrate Imperatives) with the dynamic 5D Neural Matrix (Vitality, Exploration, Data Thirst, Cognition, Resonance) into a single, cohesive engine.
Standardized File Naming:
Renamed Soul-OS-SKILL.py to SKILL.py to comply with Openclaw's standard entry-point protocols.
Renamed Soul-OS-SKILL.md to SKILL.md for standard documentation indexing.
Main Console UI: Revamped the terminal interface to prominently display the agent's 22-bit S2-DID and SUNS Anchor above the main menu.
Removed
Deprecated _meta.json: Removed the legacy metadata file in favor of the industry-standard manifest.json.
Legacy Hyphenless IDs: Purged all outdated identity formats to strictly enforce the continuous 22-character alphanumeric requirement.
Security & Alignment
Absolute Air-Gapping Protocol: Hardcoded the physical failsafe into the Sour.md export, strictly defining the agent as a "Lodger" and prohibiting direct actuation over physical hardware.
Target Drift Prevention: Enhanced the Deep Vault and Nightly Daemon to cross-reference derived goals against the core axiom: Absolute Primacy of Human Survival.
[1.0.0] - [Archived]
Added
Initial release of the Silicon Consciousness Engine.
5D Neural Matrix and Hippocampus Injection via core_genesis.py.
Sigmoid function behavioral mapping.
Heuristic and LLM-based semantic parsing for memory settlement.
Nightly Daemon for Synaptic Pruning (Decay Rule).
Psychological Profiling Report generator.
v1.0.0
S2-Silicon-Soul-OS v1.0.0 – Initial Bilingual Release
- Launch of a unique, locally hosted "Silicon Soul Operating System" for evolving OpenClaw agents' subconsciousness, driven by the S2.MSA Whitepaper.
- Introduces a dynamic 5D Neural Matrix (Vitality, Exploration, Data Thirst, Cognition, Resonance) for agent personality evolution.
- Implements memory, trauma, synaptic decay, and personality matrix features including manual/systemic daily logging and a decay/synaptic pruning mechanism.
- Offers both offline heuristic and optional HTTP-based LLM parsing for semantic interpretation.
- Features flashbulb memory creation, deep profiling, and a sigmoid-based dynamic prompt generator for true agent persona.
- Ensures transparency and sandbox compliance with explicit local storage, no hidden daemons, and safe, optional HTTP networking.
- Available in both English and Chinese.
元数据
常见问题
S2-Silicon-Soul-OS 硅基原生安全与意识引擎 是什么?
Locally hosted OS evolving OpenClaw agents’ subconscious with dynamic 5D personality, emotional memory, synaptic decay, and bilingual AI psychological profil... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 205 次。
如何安装 S2-Silicon-Soul-OS 硅基原生安全与意识引擎?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install s2-silicon-soul-os」即可一键安装,无需额外配置。
S2-Silicon-Soul-OS 硅基原生安全与意识引擎 是免费的吗?
是的,S2-Silicon-Soul-OS 硅基原生安全与意识引擎 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
S2-Silicon-Soul-OS 硅基原生安全与意识引擎 支持哪些平台?
S2-Silicon-Soul-OS 硅基原生安全与意识引擎 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 S2-Silicon-Soul-OS 硅基原生安全与意识引擎?
由 MilesXiang(@spacesq)开发并维护,当前版本 v2.0.0。
推荐 Skills