Smart writing workspace: RSS feeds, AI topic generation, and blog creation in one place.

中文文档

• What is OpenJarvis?
• Latest Updates
• Key Features
• Get Started
• Docker Deployment
• Launch from Source
• Tech Stack
• Configuration
• License

OpenJarvis is an intelligent writing assistant that combines RSS subscription, AI-powered topic generation, and LangGraph-driven blog creation. Subscribe to feeds, generate blog ideas from news, and write articles with AI assistance—all in one workflow.

• 2026-03-09 Scheduler: Replaced APScheduler with croniter-based polling (30s); reliable daily digest trigger; startup catchup when cron time already passed.
• 2026-03-08 Article quality: LLM temperature per task (0.2 for JSON, 0.7 for creative); style/audience inference (LLM + rule fallback); prompt packs (tutorial/paper_review/business); LangGraph pipeline infer→fetch→synthesize→outline→plan→write_sections→assemble→fact_check→style_polish→quality_gate; fetch uses trafilatura + cache; topic dedup; frontend "自动推断" for style/audience.
• 2026-03-07 Scheduler fix: BackgroundScheduler now initialized with TIMEZONE so cron fires at the correct local time regardless of container timezone (was UTC by default).
• 2026-03-07 title_zh split: original English title preserved in title; Chinese translation stored separately in title_zh; keyword filtering matches both; backfill skips already-translated items; summary translation removed.
• 2026-03-06 Feishu push: Digest push to Feishu (like email)—supports custom bot webhook and Flow webhook; FEISHU_WEBHOOK_URL in .env (semicolon-separated); POST /subscribe/feishu API; bilingual titles 中文 (English); rich formatting (sections, related articles); scripts/test_feishu_push.py and scripts/test_email_push.py.
• 2026-03-06 Email: HTML digest with gradient header, section blocks, topic cards; title_zh for bilingual display.
• 2026-03-06 Article Interpret: POST /article/interpret/{articleId}; structured interpretation cached in DB (interpret_result); ?force=1; friendly error when site blocks crawling.
• 2026-03-06 History: pagination; IdeaGenerator hidden on history tab; bilingual title display.
• 2026-03-06 Scheduler: Cron uses TIMEZONE; scheduler in separate container; digest checks DB first.
• 2026-03-06 RSS: FeedParser UA; updated feed URLs; scripts/diagnose_feeds.py.
• 2025-03-05 AI RSS: RSS aggregation, RSSHub, Medium config; startup pre-fetch (15s after boot, disable via STARTUP_PREFETCH_ENABLED=false); decouple /today from auto-fetch; crawler lock + StaleDataError handling; init_db no longer creates DB; migration 005.
• 2025-03-05 Docker: backend + frontend only (no postgres); user manages DB separately; removed default RSS feeds and article domains; fixed oc_conversations IntegrityError, api.ts batchCreateFeeds type.
• 2025-03-04 Unified HTTP client for RSS fetch and fetch_url; proxy support (RSS_USE_PROXY, socks5h); feed create returns 200 when already exists; fixed logging middleware crash.
• 2025-03-04 RSS create/update supports fetchNow; batch import feeds; Demo GIF.
• 2025-03-03 Docker deployment support.
• 2025-03-02 Feeds management page and FeedManager component.
• 2025-03-01 Translation support (MT / LLM).
• 2025-02-28 LangGraph workflow for blog creation.
• 2025-02-25 Email push (Resend) and Feishu Webhook integration.

• RSS Feeds: Multi-source subscription, keyword filtering, scheduled fetching
• AI Topics: Auto-generate blog ideas from news
• Smart Writing: LangGraph-driven flow (outline, drafting, quality validation)
• Push & Share: Email push (Resend), Feishu digest push (bot/Flow webhook)

• Python 3.10+
• Node.js 18+ (frontend)
• PostgreSQL 14+
• Docker & Docker Compose (for Docker deployment)

1. Start PostgreSQL (user-managed). Docker Compose does not include postgres. Run it locally or in a separate container. Ensure the DB exists and is reachable.

2. Create .env (required):

   cp .env.example .env

   Configure at minimum:

   • POSTGRES_HOST: DB host. Use host.docker.internal if postgres runs on your host machine (Mac/Win/Linux with Docker 20.10+).
   • POSTGRES_PORT, POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD: Match your postgres.
   • AI_API_KEY: AI service key
   • Email push: RESEND_API_KEY, RESEND_FROM (register at Resend)

1. Clone the repo:

   git clone https://github.com/your-org/OpenJarvis.git
   cd OpenJarvis

2. Configure .env per "Before Starting" above

3. Start services:

   docker compose up -d

   Services: backend (API, 4 workers), scheduler (RSS cron + digest), frontend.

4. Backend: http://localhost:12135
   API docs: http://localhost:12135/docs
   Frontend: http://localhost:5173 (included in docker compose up)

5. If backend fails: docker compose logs backend

cp .env.example .env
# Edit .env: POSTGRES_PASSWORD, AI_API_KEY

python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

./scripts/setup.sh

./start.sh                    # Backend http://localhost:12135
cd web && npm install && npm run dev   # Frontend http://localhost:5173

See .env.example. Main variables:

MIT