Daily Roleplay Game

Daily profession roleplay game engine with hidden kink guessing, AI-driven personality generation, achievement tracking, and multi-backend image generation (...

Security Scan

The skill's files, runtime instructions, and assets match its stated purpose (an automated daily NSFW roleplay generator) — nothing obviously misdirected — but it requires creating a persistent workspace, scheduling tasks, and later configuring messaging/image-backend credentials, so review the install scripts and consider isolating it before use.

Purpose & Capability

The name/description (daily profession roleplay with hidden-kink guessing and multi-backend image generation) align with the repository contents: large kink/profession/personality data sets, ComfyUI/SD/Midjourney templates, prompt files, and generator/engine docs. Files and templates for image backends, prompts, hint systems, and achievement tracking are expected for this purpose.

Instruction Scope

The SKILL.md instructs running scripts/setup.sh to create a persistent workspace (~/.openclaw/workspace-role-play), an autonomous 'role-play' agent, cron entries (6:00 and 23:30), and to configure USER.md, MEMORY.md and TOOLS.md (messaging channels and image backends). Those actions are consistent with the skill's goal, but they give the skill access to local files, scheduled execution, and the ability to send messages and generated images to external platforms. SKILL.md also mentions reading calendar/todo summaries for calendar-linked theme weighting — that implies access to user calendar data if you enable that integration.

Install Mechanism

There is no external install spec (no downloads or package installs). The skill is instruction-only but includes local scripts (setup.sh, validate-generation.sh, wrapup.sh) that will copy files and write to ~/.openclaw/workspace-role-play and create cron tasks. This is a standard local-install pattern; risk is limited to what those scripts do on your machine — inspect them before running.

Credentials

Registry metadata lists no required env vars, but runtime configuration expects you to provide messaging platform credentials (Discord/Telegram/Feishu/etc.), image-backend access (ComfyUI/SD WebUI endpoints or Midjourney account), and possibly calendar access. Those credentials are necessary for features the skill advertises; the skill did not declare them as required in registry metadata, so you'll need to supply them manually in your OpenClaw workspace. Only provide tokens/keys to services you trust and prefer creating dedicated service accounts where possible.

Persistence & Privilege

The setup creates a persistent workspace and an autonomous 'role-play' agent and expects cron scheduling for automated runs (6:00 and 23:30). While persistent agents and scheduled jobs are necessary for a daily game, that persistence combined with automatic posting to external channels increases the blast radius if something goes wrong. Confirm cron entries and agent configuration before enabling autonomous operation.

Assessment

This skill is internally consistent with its stated purpose (an automated NSFW roleplay generator), but it makes persistent changes and expects you to supply messaging/image-backend credentials and optionally calendar access. Before installing: 1) Inspect scripts/setup.sh, validate-generation.sh, and wrapup.sh to confirm they do only the file/cron/agent operations you expect. 2) Do not run setup.sh as an admin/root user; run in a dedicated user account or sandbox/container. 3) Only supply messaging tokens or image-backend credentials you trust (prefer dedicated/test accounts). 4) Be aware the skill will generate and store explicit images and archives under your home directory; check/archive policies and legal constraints. 5) If you plan to enable autonomous posting, verify cron settings and the configured channels to avoid accidental public posting. If you want extra assurance, request the full contents of the three scripts for review before running them.

Daily Profession Roleplay Game

AI 驱动的每日职业角色扮演系统。每天自动抽取职业、年龄、性格（五维）、隐藏性癖（4~6 个），通过三级暗示系统引导猜测，支持多种生图后端（ComfyUI / SD WebUI / Midjourney / Nano Banana Pro）。

安装

方式 1：ClawHub（推荐）

clawhub install daily-roleplay-game

安装后运行部署脚本（自动创建独立的 role-play agent，不影响现有 agent）：

./skills/daily-roleplay-game/scripts/setup.sh

方式 2：Git Clone

git clone https://github.com/nannyu/openclaw-role-play-skill.git
cd openclaw-role-play-skill
./scripts/setup.sh

脚本会自动：

1. 创建 role-play agent（如 openclaw CLI 可用）
2. 部署引擎文件（ENGINE.md, AGENTS.md, HEARTBEAT.md, SOUL.md）到 ~/.openclaw/workspace-role-play/
3. 复制数据目录（professions, kinks, themes, personality, weights, templates）
4. 从模板初始化运行时文件（MEMORY.md, USER.md, TOOLS.md, IDENTITY.md 等）
5. 创建 archive/ 和 scripts/ 目录

也可指定自定义 workspace 路径：./scripts/setup.sh /path/to/workspace

部署后配置

cd ~/.openclaw/workspace-role-play

1. IDENTITY.md — 填写角色名称和时区
2. USER.md — 填写你的个人信息
3. MEMORY.md — 配置消息频道（discord/telegram/feishu/last）
4. TOOLS.md — 配置生图工具（ComfyUI/SD WebUI/Midjourney/Nano Banana Pro，不需要可填「无」）
5. openclaw.json — 将 openclaw.example.json5 的配置合并进去（心跳、频道绑定等）
6. 定时任务 — 添加 6:00 初始化 + 23:30 收尾归档（见 docs/CRON_CONFIG.md）

完整部署指南见 docs/OPENCLAW_SETUP.md。

System Architecture

~/.openclaw/workspace-role-play/
├── SOUL.md / ENGINE.md / AGENTS.md / HEARTBEAT.md  ← 核心（静态）
├── USER.md / MEMORY.md / TOOLS.md / IDENTITY.md    ← 用户信息（手动维护）
├── roleplay-active.md      ← 每日生成（YAML front-matter + 强制模板）
├── guess-log.md             ← 当日猜测进度
├── kink_game_enabled.json   ← 玩法开关
├── data/                    ← 数据库
│   ├── professions/*.yaml   ← 13 类 ~138 个职业
│   ├── kinks/category_[a-f].yaml ← 6 类 131 个性癖
│   ├── themes/              ← 主题日配置
│   ├── personality/         ← 五维性格生成
│   ├── weights/             ← 职业+年龄性癖加权
│   └── templates/           ← 早安模板 + 生图配置
├── scripts/
│   ├── wrapup.sh            ← 23:30 收尾归档
│   └── validate-generation.sh ← 生成器输出验证
└── archive/                 ← 历史存档

Daily Flow

6:00 — 自动初始化（ENGINE.md Step 0-8）

1. 前置检查（生图工具、残留清理、re-roll 规则）
2. 抽取职业 → 主题日 → 年龄 → 性癖（3~5+1，含职业+年龄加权）→ 稀有替换
3. 生成五维性格（职业维度/自我/本我/超我/NSFW性格）
4. 写入 roleplay-active.md（强制模板，含 YAML front-matter）
5. 生成 bio.md (~800字) + personality.md (~500字) 到存档
6. 创建 guess-log.md + kink_game_enabled.json
7. 发送早安消息 + 生图自拍
8. 更新 history_tracker.json（全部四项追踪）
9. 执行 validate-generation.sh 验证输出

运行时 — Agent 行为

• 按 AGENTS.md 启动顺序读取文件，进入角色
• 猜性癖玩法默认隐藏，用户发送口令解锁
• 三级暗示：6-12 Lv.1 / 12-18 Lv.2 / 18-24 Lv.3
• 猜对脱衣+拍照，猜错3次穿回，通关全脱+惩罚照
• 禁止性癖信息泄漏

23:30 — 自动收尾（wrapup.sh）

归档 roleplay-active.md + guess-log.md + 图片 → archive/YYYY-MM-DD-职业名/

Key Files Reference

File	Purpose	Update
engine/ENGINE.md	运行时规则唯一权威 + 生成器操作手册	静态
engine/AGENTS.md	启动顺序与行为规范	静态
engine/HEARTBEAT.md	心跳规则	静态
engine/SOUL.md	角色人格核心	可自定义
data/index.yaml	数据索引与生成逻辑	扩展时修改
data/achievements.yaml	成就系统配置	可自定义

Customization

• 角色人格：编辑 engine/SOUL.md
• 添加职业：在 data/professions/ 对应分类 YAML 中追加
• 添加性癖：在 data/kinks/category_*.yaml 中追加，更新 data/index.yaml count
• 添加主题日：编辑 data/themes/daily_themes.yaml
• 生图工具配置：编辑 templates/TOOLS.md（ComfyUI 详细配置见 data/templates/comfyui/README.md）

Design Documents

详细设计文档见 docs/daily-roleplay-game.md