https://github.com/zack-dev-cm/trendwatch

Trendwatch YouTube Shorts pipeline and a small MCP server
https://github.com/zack-dev-cm/trendwatch

mcp-server youtube-api-v3 youtube-bot

Last synced: about 2 months ago
JSON representation

Trendwatch YouTube Shorts pipeline and a small MCP server

Host: GitHub
URL: https://github.com/zack-dev-cm/trendwatch
Owner: zack-dev-cm
Created: 2025-07-18T17:06:24.000Z (11 months ago)
Default Branch: main
Last Pushed: 2025-07-18T19:54:28.000Z (11 months ago)
Last Synced: 2025-07-18T21:53:35.626Z (11 months ago)
Topics: mcp-server, youtube-api-v3, youtube-bot
Language: Jupyter Notebook
Homepage: https://trendwatch-1095464065298.us-east1.run.app
Size: 1000 KB
Stars: 0
Watchers: 0
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

# MCP Connector Integration Guide

*(Trendwatch — July 2025)*

---

## 1 · Live Server Endpoints

| Purpose | URL |
| -------------------------------- | -------------------------------------------------------------------------------------------------------------- |
| **Web UI** (minimal search page) | [https://trendwatch-1095464065298.us-east1.run.app](https://trendwatch-1095464065298.us-east1.run.app) |
| **MCP SSE Endpoint** | [https://trendwatch-1095464065298.us-east1.run.app/sse](https://trendwatch-1095464065298.us-east1.run.app/sse) |

> The server is deployed on Google Cloud Run from the included `Dockerfile`. If you’re self‑hosting, replace the URLs above with your own domain.

---

## 2 · Prerequisites

* Deployed **Trendwatch** FastMCP server (see `Dockerfile` + `app/server.py`).
* Publicly‑reachable **HTTPS** URL (Cloud Run or any reverse proxy).
* (Optional) **Bearer token** set via `API_TOKEN` env var for authenticated access.

---

## 3 · Connecting From Popular AI Chat Providers

### 3.1 · OpenAI ChatGPT (Deep Research Mode)

1. Open **⋯ Settings → Beta features → Connectors** and click **➕ Add connector**.
2. Paste the following JSON, then save:

```jsonc
{
"server_label": "Trendwatch Shorts",
"server_url": "https://trendwatch-1095464065298.us-east1.run.app",
"allowed_tools": ["search", "fetch"],
"require_approval": "never" // or "per-search" if you want manual approval
}
```

![Connecting ChatGPT MCP](app/ui/static/mcp_chatgpt.png "Connecting ChatGPT MCP")
3. Start a new chat, select **Deep Research** model (“o3‑mini‑deep‑research” or newer) and pick **Trendwatch Shorts** in the *Tools* dropdown.
4. Test with: **“Find shorts about AI makeup tutorials.”**
5. Inspect the trace link to confirm tool calls succeed.

> **Tip:** Add `"require_approval": "auto"` to force silent calls while still recording traces.

#### Programmatic Usage

```python
from openai import OpenAI
client = OpenAI()
resp = client.responses.create(
model="o3-mini-deep-research",
tools=[{
"type": "mcp",
"server_label": "trendwatch",
"server_url": "https://trendwatch-1095464065298.us-east1.run.app",
"allowed_tools": ["search", "fetch"],
"require_approval": "never",
}],
input=[{"role": "user", "content": [{"type": "input_text", "text": "Top skateboarding shorts uploaded this week"}]}],
)
print(resp.choices[0].message.content[0].text)
```

---

### 3.2 · Anthropic Claude 3 (Messages API)

> **Claude 3 Opus** and later support MCP via the `anthropic-beta: mcp-client-2025-04-04` header.

```bash
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-beta: mcp-client-2025-04-04" \
-H "content-type: application/json" \
-d '{
"model": "claude-3-opus-2025-07-15",
"system": "You are a research assistant.",
"tools": [{
"type": "mcp",
"server_url": "https://trendwatch-1095464065298.us-east1.run.app",
"allowed_tools": ["search", "fetch"]
}],
"messages": [{"role":"user","content":"Give me viral cat shorts from the past 48 hours"}]
}'
```

Anthropic automatically infers when to call `search`/`fetch` based on your prompt.

---

### 3.3 · ElevenLabs 11.ai (Voice & Chat Agents)

1. Open **Dashboard → Agents → Tools (Beta)**.
2. Click **Add MCP Server**, then enter:

* **Label:** `Trendwatch`
* **Server URL:** `https://trendwatch-1095464065298.us-east1.run.app`
* **Transport:** `SSE`
* **Allowed tools:** `search, fetch`
3. Attach the toolset to your voice agent and deploy.
4. Speak: *“Tell me about trending Pokémon shorts.”* 11.ai converts the request to text, triggers the MCP calls, and returns spoken results.

> ElevenLabs limits MCP calls to 30 req/min per agent. Use Cloud Run’s concurrency settings to scale accordingly.

---

## 4 · Testing & Troubleshooting

| Symptom | Likely Cause | Fix |
| -------------------------- | ------------------------------------ | ------------------------------------------------------------ |
| **`401 Unauthorized`** | Missing/incorrect `API_TOKEN` header | Supply `Authorization: Bearer ` or unset `API_TOKEN`. |
| **`ToolNotFound: search`** | Connector allowed‑tools mismatch | Ensure `allowed_tools` includes both `search` and `fetch`. |
| **Network timeout** | Cloud Run cold start or egress block | Warm the service; check VPC egress rules. |
| **Empty result list** | Query too narrow | Confirm index contains matching titles/descriptions. |

Check the raw endpoint manually:

```bash
curl "https://trendwatch-1095464065298.us-east1.run.app/search?query=ai"
```

---

## 5 · Security Checklist

* **Input sanitization** – FastMCP escapes HTML; still validate prompt‑generated IDs.
* **Rate limiting** – Apply Cloud Armor or FastMCP’s built‑in limits.
* **Token scopes** – Issue least‑privilege API keys (read‑only access to this corpus).
* **Prompt‑injection review** – Monitor traces for attacker‑controlled instructions.
* **HTTPS everywhere** – Cloud Run terminates TLS; do **not** expose plain HTTP.

---

## 6 · Local Development Recap

```bash
# 1 – Install deps
pip install -r requirements.txt

# 2 – Run the pipeline & create a parquet
python -m app.pipeline --query "AI" --days 5 --out trend.csv

# 3 – Serve locally (SSE on :8000)
DATA_PATH=trend.csv python -m app.server
```

If `DATA_PATH` is unset or the file is missing, the server will create a small
example dataset automatically.

Then register `http://localhost:8000` as an MCP server in your test provider.

---

## 7 · References & Further Reading

* **FastMCP Docs:** [https://github.com/jlowin/fastmcp](https://github.com/jlowin/fastmcp)
* **OpenAI Cookbook – MCP examples:** [https://cookbook.openai.com/examples/partners/mcp_powered_voice_agents/mcp_powered_agents_cookbook](https://cookbook.openai.com/examples/partners/mcp_powered_voice_agents/mcp_powered_agents_cookbook)
* **Building MCP Servers for Deep Research (PDF)** – see repo docs.

---

ecosyste.ms

Data

Tools

Indexes

Applications

Experiments

Awesome

https://github.com/zack-dev-cm/trendwatch

Awesome Lists containing this project

README