AI Skill Hub 推荐使用:Paperless-iq 是一款优质的Agent工作流。AI 综合评分 7.5 分,在同类工具中表现稳健。如果你正在寻找可靠的Agent工作流解决方案,这是一个值得深入了解的选择。
Paperless-iq 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
Paperless-iq 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 方式一:pip 安装(推荐)
pip install paperless-iq
# 方式二:虚拟环境安装(推荐生产环境)
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install paperless-iq
# 方式三:从源码安装(获取最新功能)
git clone https://github.com/knows-cloud/paperless-iq
cd paperless-iq
pip install -e .
# 验证安装
python -c "import paperless_iq; print('安装成功')"
# 命令行使用
paperless-iq --help
# 基本用法
paperless-iq input_file -o output_file
# Python 代码中调用
import paperless_iq
# 示例
result = paperless_iq.process("input")
print(result)
# paperless-iq 配置文件示例(config.yml) app: name: "paperless-iq" debug: false log_level: "INFO" # 运行时指定配置文件 paperless-iq --config config.yml # 或通过环境变量配置 export PAPERLESS_IQ_API_KEY="your-key" export PAPERLESS_IQ_OUTPUT_DIR="./output"
Open-source AI layer for Paperless-NGX. Paperless IQ connects to your existing paperless-ngx instance and adds LLM-driven automatic metadata tagging, a RAG-powered conversational document search, and a background automation engine — all self-hosted, all with a human-in-the-loop approval workflow.
Works with local models via Ollama, Amazon Bedrock (Claude, Nova, Llama, Mistral, Titan), Anthropic, and OpenAI.
---
---
| Variable | Purpose |
|---|---|
PAPERLESS_URL | Base URL of the Paperless-NGX instance (internal, e.g. http://webserver:8000) |
PAPERLESS_TOKEN | API token for Paperless-NGX |
uv sync
uv run uvicorn backend.main:app --reload
docker compose build paperless-iq && docker compose up -d paperless-iq
---
Add to your Paperless-NGX docker-compose.yml:
paperless-iq:
build:
context: /path/to/paperless-iq
dockerfile: docker/Dockerfile
restart: unless-stopped
depends_on:
- webserver
ports:
- "8082:8080"
volumes:
- paperless-iq-data:/data
environment:
PAPERLESS_URL: http://webserver:8000
PAPERLESS_TOKEN: <your-paperless-api-token>
SECRET_KEY: <random-secret-for-encryption>
# Optional: pre-configure LLM on first run
PIQ_LLM_PROVIDER: bedrock
PIQ_LLM_MODEL: eu.anthropic.claude-haiku-4-5-20251001-v1:0
Add to the volumes: section:
volumes:
paperless-iq-data:
Then:
docker compose up -d --build paperless-iq
Access the UI at http://localhost:8082.
---
Settings are organised into eight tabs:
| Tab | Contents |
|---|---|
| **Connection** | Paperless-NGX public URL, connection test, inbox tag, webhook registration |
| **AI Provider** | LLM provider + model + credentials, context window, analysis mode, embedding provider, vector store backend |
| **Prompts & Fields** | Global system prompt, LLM output language, per-field instructions, custom fields |
| **Metadata Rules** | Smart entity selection toggle, similar-docs count, frequency fallback, entity creation policies |
| **Automation** | Enable/disable, auto-apply, poll interval, batch size, cron schedule, creation policies |
| **Appearance** | Theme colours, typography, logo, nav icons, UI language, colour scheme (light/dark/auto) |
| **Memories** | Enable/disable long-term memory, list/edit/delete individual facts, clear all |
| **Access Control** | Per-user permission flags, NG admin sync toggle, maintenance actions (reindex, reset tracking) |
All settings are configurable via the web UI. On first startup, settings can be seeded from environment variables (prefixed PIQ_). After the first UI save, database values take precedence over environment variables.
| Variable | Default | Purpose |
|---|---|---|
CORS_ALLOWED_ORIGINS | * | Comma-separated list of allowed CORS origins. Restrict this in production (e.g. https://paperless.example.com). |
Webhook secret — Paperless IQ auto-generates a webhook secret on first startup and embeds it in the callback URL registered with Paperless-NGX. No manual configuration is required.
| Variable | Default | Purpose |
|---|---|---|
PIQ_LLM_PROVIDER | ollama | ollama · anthropic · openai · bedrock |
PIQ_LLM_MODEL | llama3 | Model name (provider-specific) |
PIQ_LLM_CREDENTIALS | — | API key (Anthropic/OpenAI) or JSON credentials (Bedrock) |
PIQ_OLLAMA_URL | http://localhost:11434 | Ollama server URL |
PIQ_EMBED_PROVIDER | ollama | Embedding provider: ollama · openai · bedrock |
PIQ_EMBEDDING_MODEL | nomic-embed-text | Embedding model name |
PIQ_DEFAULT_ANALYSIS_MODE | ocr | ocr or full_document |
PIQ_CONTEXT_WINDOW_CHARS | 128000 | Max characters sent to LLM per request |
PIQ_SMART_ENTITY_SELECTION | true | Use vector similarity for entity pre-selection |
PIQ_SIMILAR_DOCS_COUNT | 10 | Similar documents to retrieve for entity selection |
PIQ_FREQUENCY_FALLBACK_COUNT | 20 | Top-N frequent entities used as fallback |
PIQ_TAG_CREATION_POLICY | existing_only | existing_only or allow_new |
PIQ_CORRESPONDENT_CREATION_POLICY | existing_only | existing_only or allow_new |
PIQ_DOCTYPE_CREATION_POLICY | existing_only | existing_only or allow_new |
PIQ_INBOX_TAG_ID | — | Paperless-NGX tag ID for the inbox |
PIQ_AUTO_APPLY | false | Skip the approval queue |
PIQ_AUTOMATION_ENABLED | false | Enable inbox polling and scheduled runs |
PIQ_POLL_INTERVAL_SECONDS | 10 | Inbox poll interval |
PIQ_BATCH_SIZE | 10 | Documents per scheduled batch |
PIQ_SCHEDULE_CRON | — | Cron expression for batch runs |
PIQ_AUDIT_RETENTION_DAYS | 90 | Days before audit entries are pruned |
PIQ_TARGET_LANGUAGE | — | Language for LLM responses (e.g. German) |
PIQ_VECTOR_STORE_BACKEND | local | local (ChromaDB) or bedrock_kb |
PIQ_BEDROCK_KB_ID | — | Bedrock Knowledge Base ID |
PIQ_MEMORY_ENABLED | true | Enable long-term memory extraction |
---
高质量的自动化工作流工具
该工具使用 NOASSERTION 协议,商用场景请仔细阅读协议条款,必要时咨询法律意见。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
📄 NOASSERTION — 请查阅原始协议条款了解具体使用限制。
总体来看,Paperless-iq 是一款质量良好的Agent工作流,在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态,建议收藏备用,结合自身场景选择合适时机引入使用。
| 原始名称 | paperless-iq |
| 原始描述 | 开源AI工作流:Open-source AI add-on for Paperless-NGX: automatic metadata & tagging, RAG docum。⭐7 · Python |
| Topics | aiautomationpython |
| GitHub | https://github.com/knows-cloud/paperless-iq |
| License | NOASSERTION |
| 语言 | Python |
收录时间:2026-05-29 · 更新时间:2026-05-30 · License:NOASSERTION · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端