mongodb-query - SKILL.md Agent Skill

name: mongodb-query description: Query MongoDB notes store for memory analysis and statistics. triggers: - mongodb - mongo - notes store - memory storage - notes query

MongoDB Query Skill

Query the MongoDB notes store to investigate memory contents, embeddings, and note statistics.

Connection Details

URI: mongodb://localhost:27017 (or MONGODB_URI env var)
Database: qq_memory
Collection: notes

Python Usage

from qq.memory.mongo_store import MongoNotesStore

# Initialize store
store = MongoNotesStore()

# Get a specific note
note = store.get_note("note_id_here")

# Get recent notes
recent = store.get_recent_notes(limit=10)

# Get notes by importance range
important = store.get_by_importance_range(min_importance=0.7, max_importance=1.0)

# Get stale notes (not accessed recently)
stale = store.get_stale_notes(days_threshold=30)

Direct PyMongo Usage

from pymongo import MongoClient

client = MongoClient("mongodb://localhost:27017")
db = client["qq_memory"]
notes = db["notes"]

# Count all notes
total = notes.count_documents({})

# Find all notes
all_notes = list(notes.find({}, {"content": 1, "section": 1, "importance": 1}))

# Count by section
pipeline = [
    {"$group": {"_id": "$section", "count": {"$sum": 1}}},
    {"$sort": {"count": -1}}
]
by_section = list(notes.aggregate(pipeline))

# Find notes without embeddings
no_embedding = notes.count_documents({"embedding": None})

# Sample notes
sample = list(notes.find().limit(10))

CLI Usage

# Use mongosh directly
docker exec -it qq-mongodb-1 mongosh qq_memory --eval "db.notes.countDocuments({})"

# Get collection stats
docker exec -it qq-mongodb-1 mongosh qq_memory --eval "db.notes.stats()"

Notes Schema

Each note document contains:

note_id: Unique identifier
content: Note text
embedding: Vector embedding (list of floats)
section: Category (e.g., "Key Topics", "Preferences")
metadata: Additional key-value pairs
importance: Score 0.0-1.0 (default 0.5)
decay_rate: How fast importance decays (default 0.01)
access_count: Number of times accessed
last_accessed: Timestamp of last access
created_at: Creation timestamp
updated_at: Last update timestamp