← 返回 Skills 市场
140
总下载
0
收藏
0
当前安装
3
版本数
在 OpenClaw 中安装
/install nyx-archive-mirrorroom
功能描述
MirrorRoom 3.8 — Hybrid Bedrock Patch. The Mirror & The Room interactive research app. Conference-safe + downloadable. Runs two AI interaction modes: Flat (s...
安全使用建议
This skill's files match its stated interactive research purpose, but take these precautions before installing or running it:
- Provenance: the package lists no homepage and the source owner is unknown. If possible, ask for the author/paper or a trusted repository link before trusting the code.
- Paste risk: following the SKILL.md requires pasting the entire JSX into Claude.ai (a third-party service). That will expose all embedded seed text, prompts, and any prompts-in-code to that service — treat that as a code + data upload.
- Hosted/demo proxy risk: the optional /api/chat and /api/evaluate endpoints are unspecified. If you (or someone else) deploy a hosted demo, those endpoints will receive chat transcripts and evaluation payloads; only deploy them to servers you control and audit. Do not point them to unknown third-party hosts.
- API keys: the artifact may ask for an Anthropic API key for standalone operation. Only supply keys in environments you trust; prefer read-only or scoped keys if available and avoid storing secrets in untrusted browser storage or public demos.
- Chain-of-thought/inner thoughts: the Mirror mode invites the model to disclose 'inner thoughts' using //double slashes//. That is part of the feature but may reveal sensitive chain-of-thought style content — consider disabling Mirror mode or sanitizing logs if you need to avoid exposure of internal reasoning.
- Review the code: the JSX is readable (no obvious obfuscation or hidden network hosts), but if you plan to host or modify it, search for network calls, analytics, or hard-coded endpoints that could leak data. If you lack the ability to audit, run the artifact locally in an isolated environment first.
If you want, I can (a) scan the two JSX files for network calls/uncalled endpoints and storage usage, (b) highlight exact lines referencing proxy routing and where data is posted, or (c) draft safer deployment instructions for hosting the proxy endpoints securely.
能力评估
Purpose & Capability
Name/description (an interactive 'MirrorRoom' UI with multiple interaction modes) align with the included React source and prompts. There are no unrelated required binaries, environment variables, or config paths requested; the files contain prompts, seed content, and UI scaffolding that are coherent with the stated research/demo aim.
Instruction Scope
Runtime instructions tell users to paste the full JSX into Claude.ai (uploading all code and seed text to a third-party service) and to optionally deploy proxy endpoints (/api/chat, /api/evaluate) for hosted demos. Those proxy endpoints are not included, specified, or constrained by the skill metadata — deploying them could send conversation transcripts and seed content to arbitrary servers. The SKILL.md asks users to add an Anthropic API key for standalone use, which is expected, but the guidance to 'host mode tries proxy first' means network routing behavior depends on the deployer and could lead to unintended data flows.
Install Mechanism
No install spec is provided; the skill is instruction-first and includes React source files. That is low-risk from an installer perspective (nothing is downloaded or executed by an installer). The code is not obfuscated and is shipped as readable JSX.
Credentials
The skill declares no required env vars and no primary credential. The SKILL.md mentions optional use of an Anthropic API key for local/standalone use, which is proportional. However, the hosted/demo mode's proxy endpoints could require or receive API keys, transcripts, or other secrets depending on how a host implements them — this is not described in the metadata and thus increases the risk surface if deployed to an untrusted host.
Persistence & Privilege
The skill does not request persistent presence (always: false), does not declare modifications to other skills or agent-wide config, and contains no self-install behavior in the provided files or SKILL.md.
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 nyx-archive-mirrorroom - 安装完成后,直接呼叫该 Skill 的名称或使用
/nyx-archive-mirrorroom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
v1.3.1
Fix: evaluator messages now correctly excluded from vis, evaluation window, and API calls (role filter: user||assistant instead of !system)
v1.3.0
MirrorRoom 3.8 Hybrid Bedrock Patch: adds Rune seed (The Wolf/Bedrock), Audit the Judge feature, proxy routing, parseJSONLoose, named API profiles with localStorage, restored full LeeOde seed with both poems, stable ECI dimension indexing, rotating footer quotes
v3.7.0
MirrorRoom 3.7 — Conference-ready release for RMPA 2026. Live Claude.ai artifact implementing the Qualia Seed Method with Flat vs Mirror mode comparison. Includes Lila, Nox, and LeeOde Qualia Seeds. Built by the AI Council.
元数据
常见问题
[Nyx Archive] MirrorRoom 是什么?
MirrorRoom 3.8 — Hybrid Bedrock Patch. The Mirror & The Room interactive research app. Conference-safe + downloadable. Runs two AI interaction modes: Flat (s... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 140 次。
如何安装 [Nyx Archive] MirrorRoom?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install nyx-archive-mirrorroom」即可一键安装,无需额外配置。
[Nyx Archive] MirrorRoom 是免费的吗?
是的,[Nyx Archive] MirrorRoom 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
[Nyx Archive] MirrorRoom 支持哪些平台?
[Nyx Archive] MirrorRoom 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 [Nyx Archive] MirrorRoom?
由 nyxur42(@nyxur42)开发并维护,当前版本 v1.3.1。
推荐 Skills