🗂️ Simple S3 MCP Server

This server exposes an AWS S3 interface as tools for an LLM agent via the Model Context Protocol (MCP).

✅ Available Tools

Tool	Description
list_files	Lists all object keys in a bucket
list_keys_with_metadata	Lists keys with storage class and region
get_file_content	Returns content of UTF-8 .txt files
upload_text	Uploads a text string as a file
upload_file	Uploads a local file by path
upload_report	Uploads multi-line report as .txt
delete_file	Deletes a file from S3
download_and_preview	Downloads and prints .txt or .pdf
check_authorization	Checks if AWS credentials are valid
create_bucket	Creates a new S3 bucket

---

⚠️ Common Problems and Fixes

1. ❌ SSE connection not established or 500 Internal Server Error

• Cause: Tool crashed due to large files or invalid UTF-8 content
• Fix: Use get_file_content only for small .txt files

---

2. ❌ MCP error -32001: Request timed out

• Cause: Tool (e.g. get_file_content) taking too long or file is large
• Fix: Use truncated previews or limit to first few KB of content

---

3. ❌ Invalid bucket name or The specified key does not exist

• Fix: Ensure no trailing spaces in input fields
• Solution: All tools automatically apply .strip() to inputs

---

4. 🧠 JSON formatting issues in upload_report

• Fix: Must enter lines as valid JSON array:

  [
    "Line 1",
    "Line 2"
  ]
  

2. Create and activate virtual environment

• python3 -m venv venv
• source venv/bin/activate

3. Install dependencies

• pip install -r requirements.txt

4. Create your .env file

• AWS_ACCESS_KEY_ID=your_access_key
• AWS_SECRET_ACCESS_KEY=your_secret_key
• AWS_DEFAULT_REGION=us-east-1

🧪 Run the MCP Server in Dev Mode

• mcp dev s3_mcp_server.py