← 返回 Skills 市场
Persona Builder
作者
Corbin Breton
· GitHub ↗
· v2.0.0
· MIT-0
186
总下载
0
收藏
2
当前安装
3
版本数
在 OpenClaw 中安装
/install persona-builder
功能描述
Guided interview to generate a complete agent workspace: SOUL.md, IDENTITY.md, MEMORY.md, AGENTS.md, USER.md with hierarchical memory structure and atomic fa...
安全使用建议
This skill is largely documentation and templates for building a persona; nothing in the bundle requests credentials or installs code. Before trusting it in production: (1) be aware the README claims a 'persona-builder' CLI and weekly scripts (items.json, decay_sweep) but this package contains only templates and docs — the described automation is not present here; verify any separate installer/source before running. (2) Review and edit the generated files yourself — do not assume the default behavior or default autonomy settings are safe for your use case (defaults may enable aggressive push-back/autonomy if you skip questions). (3) The skill states ‘‘nothing is transmitted externally’’ but that is a documentation claim — if you later connect these templates to other tooling (Ops Toolkit, nightly extraction, sub-agents), re-check what those integrations require (API keys, network access). (4) If you need a CLI, confirm its provenance (official release or third-party) before installing. Overall this bundle is internally consistent with its stated purpose, but verify any external tooling it references before enabling automation.
功能分析
Type: OpenClaw Skill
Name: persona-builder
Version: 2.0.0
The persona-builder skill is a legitimate configuration utility designed to initialize an OpenClaw agent's workspace. It uses a structured interview process to populate local Markdown templates (SOUL.md, IDENTITY.md, etc.) and a JSON fact store (items.json). The skill incorporates security-positive design patterns, such as 'Anti-Sycophancy Rules' and 'Epistemic Standards,' which are intended to reduce AI hallucination and improve resistance to user manipulation. No evidence of data exfiltration, malicious command execution, or unauthorized persistence was found across the code or documentation.
能力评估
Purpose & Capability
Name/description (guided interview → generate SOUL.md, IDENTITY.md, MEMORY.md, AGENTS.md, USER.md) match the provided templates and mapping rules. The skill requests no unrelated credentials, binaries, or config paths.
Instruction Scope
SKILL.md and the README are comprehensive and confined to local workspace generation and template rendering. Two things to note: (1) the docs mention runtime artifacts (items.json, weekly decay_sweep, a persona-builder CLI, Ops Toolkit integrations) but no code files or install spec are provided — these behaviors are described but not implemented here; (2) several defaults (e.g., Block 3 defaults to 'Blunt and direct' and 'Always challenge' if skipped) could change agent autonomy/behavior if users skip questions and might be surprising. The skill also asserts "Nothing is transmitted externally," which is a policy claim in prose — with no implementation present there's nothing to verify or enforce.
Install Mechanism
Instruction-only skill with no install spec and no downloads — lowest-risk install posture. README references a CLI and install command, but no install files are present; this is a documentation/incoherence note, not an active install risk.
Credentials
No required environment variables, no credential requests, and templates only reference local workspace files. Requested fields are user-provided interview answers (name, timezone, goals, etc.) which are proportionate to the stated functionality.
Persistence & Privilege
always:false and no special privileges are requested. The templates discuss agent-managed memory decay and sub-agent behavior but there's no code that autonomously persists or modifies system-wide settings. If integrated later with Ops Toolkit or scripts, that could change, but not present in this package.
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 persona-builder - 安装完成后,直接呼叫该 Skill 的名称或使用
/persona-builder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
v2.0.0: Added Block 6 (Epistemic Standards) and Block 7 (Anti-Sycophancy). Universal anti-sycophancy rules are non-negotiable for all generated personas. New SOUL template includes Epistemic Standards, Dissent Protocol, and Anti-Sycophancy Rules sections. Updated generation rules and defaults.
v1.0.1
Add privacy framing for interview data
v1.0.0
Initial publish under canonical corbin-breton owner
元数据
常见问题
Persona Builder 是什么?
Guided interview to generate a complete agent workspace: SOUL.md, IDENTITY.md, MEMORY.md, AGENTS.md, USER.md with hierarchical memory structure and atomic fa... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 186 次。
如何安装 Persona Builder?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install persona-builder」即可一键安装,无需额外配置。
Persona Builder 是免费的吗?
是的,Persona Builder 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Persona Builder 支持哪些平台?
Persona Builder 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 Persona Builder?
由 Corbin Breton(@corbin-breton)开发并维护,当前版本 v2.0.0。
推荐 Skills