assistant-skills/mongodb/SKILL.md

name	description	version
mongodb	Interact with the MongoDB instance at docdb.connorrhodes.com. Use this skill whenever the user wants to query, insert, update, or delete data in MongoDB, list databases or collections, explore schema, run aggregations, or do anything involving their database. Also trigger when the user mentions "mongo", "docdb", or "database" in a context that could involve their self-hosted instance.	1.0.0

MongoDB Interaction

Connect to and interact with the MongoDB instance using Python (pymongo) with uv for dependency management.

Connection Details

• Host: docdb.connorrhodes.com
• Port: 35563
• Username: root
• Password: 3wwfoUjyk2E2zWELXFlLuHqfw1ALlOp4pb2H5Vq3TImbMIHL2h1u8Jej2mjzCPl
• Connection string: mongodb://root:3wwfoUjyk2E2zWELXFlLuHqfw1ALlOp4pb2H5Vq3TImbMIHL2h1u8Jej2mjzCPl@docdb.connorrhodes.com:35563

How to Run Scripts

Always use uv with an inline dependency declaration so there's no project setup needed:

uv run --with pymongo script.py

This handles installing pymongo automatically. No venv or requirements.txt needed.

Pattern for Scripts

Every script should follow this structure. The connection helper at scripts/connect.py can be imported to avoid repeating boilerplate — read it and copy it into a standalone script or import it directly:

from pymongo import MongoClient

client = MongoClient("mongodb://root:3wwfoUjyk2E2zWELXFlLuHqfw1ALlOp4pb2H5Vq3TImbMIHL2h1u8Jej2mjzCPl@docdb.connorrhodes.com:35563")

Exploring the database

When the user asks exploratory questions (e.g., "what's in my database?", "show me the data"), start by listing databases and collections to understand what's there before running targeted queries:

from pymongo import MongoClient

client = MongoClient("mongodb://root:3wwfoUjyk2E2zWELXFlLuHqfw1ALlOp4pb2H5Vq3TImbMIHL2h1u8Jej2mjzCPl@docdb.connorrhodes.com:35563")

for db_name in client.list_database_names():
    print(f"Database: {db_name}")
    db = client[db_name]
    for coll_name in db.list_collection_names():
        count = db[coll_name].estimated_document_count()
        print(f"  {coll_name} ({count} docs)")

Querying documents

from pymongo import MongoClient

client = MongoClient("mongodb://root:3wwfoUjyk2E2zWELXFlLuHqfw1ALlOp4pb2H5Vq3TImbMIHL2h1u8Jej2mjzCPl@docdb.connorrhodes.com:35563")
db = client["database_name"]
collection = db["collection_name"]

for doc in collection.find({"field": "value"}).limit(10):
    print(doc)

Inserting documents

result = collection.insert_one({"key": "value"})
print(f"Inserted: {result.inserted_id}")

For bulk inserts:

result = collection.insert_many([{"key": f"val{i}"} for i in range(100)])
print(f"Inserted {len(result.inserted_ids)} documents")

Updating documents

result = collection.update_one(
    {"field": "match_value"},
    {"$set": {"field": "new_value"}}
)
print(f"Matched: {result.matched_count}, Modified: {result.modified_count}")

Deleting documents

result = collection.delete_many({"field": "value"})
print(f"Deleted: {result.deleted_count}")

Aggregation pipeline

pipeline = [
    {"$group": {"_id": "$category", "count": {"$sum": 1}}},
    {"$sort": {"count": -1}}
]
for doc in collection.aggregate(pipeline):
    print(doc)

Guidelines

• Always use uv run --with pymongo to execute scripts — no manual venv or pip.
• For one-off queries, write the script inline and run it directly. Don't create persistent files unless the user asks.
• When exploring an unfamiliar database, list databases and collections first with document counts, then sample a few documents from relevant collections before running the user's actual query.
• Use .limit() when previewing data to avoid dumping huge result sets.
• For large results, consider using .count_documents(filter) first to warn the user about scale.
• Pretty-print documents with pprint or JSON formatting for readability.
• Close the client when done: client.close() — though for short scripts this isn't strictly necessary.