closedloop-technologies / jina-ai
Install for your project team
Run this command in your project directory to install the skill for your entire team:
mkdir -p .claude/skills/jina-ai && curl -L -o skill.zip "https://fastmcp.me/Skills/Download/398" && unzip -o skill.zip -d .claude/skills/jina-ai && rm skill.zip
Project Skills
This skill will be saved in .claude/skills/jina-ai/ and checked into git. All team members will have access to it automatically.
Important: Please verify the skill by reviewing its instructions before using it.
Use Jina AI APIs for converting URLs to LLM-friendly Markdown (Reader) and searching the web (Search).
1 views
0 installs
Skill Content
---
name: jina-ai
description: Use Jina AI APIs for converting URLs to LLM-friendly Markdown (Reader) and searching the web (Search).
---
# Jina AI Skill
This skill integrates Jina AI's Reader (r.jina.ai) and Search (s.jina.ai) APIs to convert URLs and web search results into clean, LLM-friendly Markdown.
## Setup
1. **Dependencies:** Requires `requests`.
```bash
pip install requests python-dotenv
```
2. **API Key Configuration (Recommended):** Jina APIs offer higher limits with an API key (JINA_API_KEY).
```bash
read -p "Enter your Jina API key (optional, press Enter to skip): " JINA_KEY
if [ ! -z "$JINA_KEY" ]; then
echo "JINA_API_KEY=$JINA_KEY" >> .env
if [ -f .gitignore ] && ! grep -q ".env" .gitignore; then echo ".env" >> .gitignore; fi
echo "API key saved to .env."
fi
```
## Usage
The script `scripts/jina_tools.py` supports `read` and `search` operations.
### 1. Read URL (Jina Reader)
Convert a URL into clean Markdown.
```bash
python3 scripts/jina_tools.py read "<url>"
```
**Example:**
```bash
python3 scripts/jina_tools.py read "https://en.wikipedia.org/wiki/Large_language_model"
```
### 2. Search Web (Jina Search)
Search the web and return the top results converted to Markdown.
```bash
python3 scripts/jina_tools.py search "<query>"
```
**Example:**
```bash
python3 scripts/jina_tools.py search "What is Jina AI?"
```
## Output
The script outputs results in JSON format, containing title, URL, and `content_markdown`.