🔌 ocplatform-pipepost

[image: ClawHub] [image: GitHub] [image: License: MIT]

Native OpenClaw plugin for PipePost — AI-powered content curation pipeline.

Scout articles from HackerNews, Reddit, RSS, and web search → translate to any language with AI → publish to your blog, Telegram, or any webhook. All from your OpenClaw agent.

What you get

5 native tools available to your OpenClaw agent:

How it works

You: "Scout HackerNews for AI articles, translate to Russian, publish to my blog"

Agent calls:
  1. pipepost_init → generates YAML flow config
  2. pipepost_validate → checks config is valid
  3. pipepost_run --dry-run → previews what it finds
  4. pipepost_run → executes the pipeline

The plugin is a thin TypeScript wrapper that calls the PipePost CLI (Python) via child process. PipePost does the heavy lifting — scouting, scoring, translating, and publishing.

Install

1. Install PipePost (Python engine)

pip install pipepost

2. Install the OpenClaw plugin

# From ClawHub
openclaw plugins install clawhub:openclaw-pipepost

# Or from GitHub
git clone https://github.com/DenSul/openclaw-pipepost
cd openclaw-pipepost && npm install && npm run build
openclaw plugins install . --dangerously-force-unsafe-install

3. Configure

Set your LLM API key:

export OPENAI_API_KEY="your-key"
export PIPEPOST_MODEL="deepseek/deepseek-chat"  # or gpt-4o, claude-sonnet, etc.

Plugin config in openclaw.json:

{
  "plugins": {
    "entries": {
      "pipepost": {
        "enabled": true,
        "config": {
          "pipepostBin": "pipepost",
          "flowsDir": "/path/to/your/flows",
          "dbPath": "/path/to/pipepost.db"
        }
      }
    }
  }
}

4. Restart OpenClaw

openclaw gateway restart

Usage Examples

Create a blog pipeline

"Create a content pipeline that scouts HackerNews for AI articles, translates to Russian, and publishes to my webhook at http://myblog.com/api/posts";

Run an existing flow

"Run the HN pipeline in dry-run mode"

Check status

"What's the PipePost status? Any articles in the database?"

Generate a new config

"Generate a flow config for Reddit programming subreddits, translate to German, output as markdown"

Flow Config Example

sources:
  - name: hackernews
    type: hackernews
    min_score: 100

translate:
  model: deepseek/deepseek-chat
  target_lang: ru

destination:
  type: webhook
  url: http://localhost:8000/api/posts/auto-publish

flow:
  steps: [dedup, scout, filter, score, fetch, translate, adapt, validate, publish, post_publish]
  filter:
    keywords_exclude: ["hiring", "ask hn"]
    min_title_length: 15
  score:
    niche: tech
    max_score_candidates: 3
  adapt:
    style: blog
  storage:
    db_path: ./pipepost.db

More examples in examples/.

Supported Sources

Supported Destinations

Requirements

• OpenClaw >= 2026.3.x
• PipePost >= 0.1.0 (pip install pipepost)
• Python >= 3.11
• Node.js >= 22 (for the plugin)
• An OpenAI-compatible LLM API key

Architecture

OpenClaw (Node.js)
    ↓ native tools
openclaw-pipepost (TypeScript plugin)
    ↓ child_process
PipePost CLI (Python)
    ↓ async pipeline
Sources → Score → Translate → Publish → Destinations

Development

git clone https://github.com/DenSul/openclaw-pipepost
cd openclaw-pipepost
npm install
npm run build     # compile TypeScript
npm run dev       # watch mode

Links

• PipePost — the Python engine
• ClawHub — plugin page
• OpenClaw — the AI assistant platform
• Example flows — ready-to-use YAML configs

License

MIT

Built by Denis Sultanov