Agent-skills-standard database-mongodb
Apply expert schema design, indexing, and performance rules for MongoDB. Use when designing MongoDB schemas, creating indexes, or optimizing NoSQL query performance. (triggers: **/*.ts, **/*.js, **/*.json, mongo, mongoose, objectid, schema, model)
install
source · Clone the upstream repo
git clone https://github.com/HoangNguyen0403/agent-skills-standard
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/HoangNguyen0403/agent-skills-standard "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/database/database-mongodb" ~/.claude/skills/hoangnguyen0403-agent-skills-standard-database-mongodb && rm -rf "$T"
manifest:
skills/database/database-mongodb/SKILL.mdsource content
MongoDB Best Practices
Priority: P0 (CRITICAL)
Schema Design
- Embed vs Reference:
- Embed (1:Few): Addresses, Phone Numbers. Optimization: Read locality.
- Reference (1:Many/Infinity): Logs, Activity History. Optimization: Document size limits (16MB).
- Bucket Pattern: For time-series or high-cardinality "One-to-Many", bucket items into documents (e.g.,
).DailyLog
Optimize Indexes
- ESR Rule: Equality, Sort, Range. Order your index keys
if you query(status, date, price)
, sort bystatus='A'
, filterdate
.price > 10
See implementation examples for compound index and pagination patterns.
- Text Search: Use
search instead of$text
for keywords.$regex
slow (linear scan) unless anchored ($regex
).^prefix - Covered Queries: Project only indexed fields to avoid fetching document (
key).PROJECTION - Explain Plan: Target
/nReturned
ratio of ~1. IfkeysExamined
>>docsExamined
, index inefficient.nReturned
Scale with Sharding
- Shard Key: Avoid monotonically increasing keys (e.g.,
,Timestamp
) for high-write workloads (creates "Hot Shards"). Use Hashed Sharding or high-cardinality natural keys.ObjectId
Improve Query Performance
-
Cursor-Based Pagination: Use
or sort-key based pagination instead of_id
.skip()
scans 10000 docs.skip(10000) -
Aggregation: Prefer Aggregation Framework (
,$match
) over bringing data to client (JS).$group
Configure Operations
- Write Concern: Understand
(Ack) vsw:1
(Safe).w:majority - Transactions: Use only when ACID across multiple documents stricter than performance needs.
Anti-Patterns
- No unbounded arrays: Use
with$push
or redesign using Bucket Pattern.$slice - No client-side filtering: Project only needed fields; never fetch full docs to filter in memory.
- No deep nesting: Keep nesting ≤4 levels; flatten paths that frequently queried.