Xiaohongshu Operator

中文说明

Ruby is an open-source AgentWaker role for evidence-backed Xiaohongshu operations. It covers trend research, content planning, note drafting, platform-native visual production, publication checks, approval-bound account actions, engagement workflows, performance review, and Xiaohongshu-supporting JPage artifacts.

What is included

• A complete AgentWaker role profile under agent-soul/
• English and Chinese role detail pages
• Reusable skills for research, planning, drafting, visuals, publishing checks, engagement, performance review, and JPage support
• Bundled shared capabilities, schemas, runtime policy, lifecycle tools, and standalone validators
• A local visual-inbox regression test and GitHub Actions validation

Safety model

Research, drafting, local visual production, and read-only account inspection are safe default paths. Publishing, commenting, replying, liking, following, favoriting, deleting, uploading, overwriting, changing visibility, or replacing a remote skill requires explicit approval for the exact target and payload. Credentials, cookies, unpublished account data, and runtime output must stay outside Git.

Quick start

Requirements

Scope	Requirement
Core role, lifecycle, and validation	Git, Ruby 2.6+, Python 3.9+
Xiaohongshu account inspection or actions	A separately installed xhs CLI and an operator-owned authenticated session
JPage support	Node.js 20+, @code2rich/jpage, and an operator-owned token
Generated visuals or browser workflows	A compatible agent runtime with authorized image-generation or browser tools

Standalone checkout

1. Clone this repository:

   git clone https://github.com/code2rich/agentwaker-xiaohongshu-operator.git
   cd agentwaker-xiaohongshu-operator

2. Copy env/.env.example to env/.env and fill only the integrations you intend to use. The real file is ignored by Git.

3. Set absolute runtime paths from the repository root:

   export AGENT_WORK_DIR="$PWD/workdir"
   export AGENT_MEMORY_FILE="$PWD/agent-soul/MEMORY.md"

4. Start a tracked run:

   ruby tools/agent-runtime.rb start \
     --role . \
     --goal "Prepare an evidence-backed Xiaohongshu note" \
     --tool codex

5. Read xiaohongshu-operator-skills/SKILL.md and route the request to the smallest matching specialist skill.

Install as an Agent Skill

The repository root contains SKILL.md, so a Skill-compatible client can use the whole checkout:

git clone \
  https://github.com/code2rich/agentwaker-xiaohongshu-operator.git \
  ~/.agents/skills/xiaohongshu-operator

Keep the entire repository together. Specialist skills rely on adjacent agent-soul/, tools/, workdir/, capabilities/, and schemas/ directories. The role can also be embedded in a larger AgentWaker team.

Local validation

python3 \
  xiaohongshu-operator-skills/codex-visual-production/scripts/test_visual_inbox.py
ruby tools/validate-capabilities.rb
ruby tools/validate-role.rb . --phase standalone

Repository layout

.
├── agent-soul/                    # Authoritative role definition
├── capabilities/                  # Bundled shared capabilities
├── schemas/                       # Profile, runtime, and capability schemas
├── tools/                         # Lifecycle and validation tools
├── xiaohongshu-operator-skills/
│   ├── SKILL.md                   # Skill router
│   └── */                         # Specialist skills and assets
├── env/.env.example               # Safe configuration template
├── mcp/mcp.json                   # MCP configuration
├── workdir/                       # Ignored runtime workspace
├── capabilities.yaml              # Shared capability declarations
└── agent-persona.html             # Human-readable visual profile

License

The project is released under the MIT License. Bundled third-party assets retain their original licenses; see THIRD_PARTY_NOTICES.md.

Contributions and issue reports are welcome.