A Kiro power that connects your IDE directly to MongoDB databases and Atlas clusters. Query collections, design schemas, optimize indexes, translate natural language to queries, and manage Atlas infrastructure — all through natural language in your editor.
MongoDB Atlas on AWS enables organizations to build intelligent, AI-powered applications that scale effortlessly. As a fully managed NoSQL database and vector search platform, Atlas unifies operational data and search in a single system, helping teams move from prototype to production with enterprise-grade security, high availability, and seamless AWS integrations. Try MongoDB Atlas (Mongo as a Service) today with the free trial tier and get 512 MB of storage at no cost. To get started, click on this link.
This power gives Kiro the ability to talk to your MongoDB deployments via the MongoDB MCP Server, and brings in expert guidance through skills adapted from the mongodb/agent-skills repository. Instead of switching between your IDE, MongoDB Compass, and the Atlas console, you can ask Kiro things like:
- "Show me all orders placed in the last 7 days with status pending"
- "Why is this aggregation pipeline slow?"
- "Should I embed reviews in my product documents or use a separate collection?"
- "Create a free Atlas cluster called staging-v2 in US East"
- "Set up vector search on my products collection"
- "Build a stream processing pipeline that reads from my Kafka orders topic"
Kiro will run the right queries, inspect your schema and indexes, and give you actionable advice grounded in 7 expert skills for MongoDB.
- Kiro IDE installed
- Node.js v20.19.0+, v22.12.0+, or v23+
- A MongoDB Atlas cluster or self-hosted MongoDB deployment
- Docker (optional, for Atlas Local development)
Open Kiro and install the MongoDB power from the Powers panel, or add it directly from the GitHub URL:
https://github.com/mongodb/kiro-powers
When prompted, configure your connection:
MDB_MCP_CONNECTION_STRING— your MongoDB connection URI (e.g.mongodb+srv://user:pass@cluster.mongodb.net/mydb)MDB_MCP_API_CLIENT_ID/MDB_MCP_API_CLIENT_SECRET— (optional) Atlas Service Account credentials for cluster management
A natural entry point is to explore your databases and collections:
> "List my databases and collections"
> "Show me the schema for the orders collection"
Then query, optimize, and build:
> "Find the top 10 customers by total spend this month"
> "Why is this query doing a COLLSCAN?"
> "Create a compound index for this filter and sort pattern"
> "Translate this question into an aggregation pipeline: revenue by region for Q1"
Query & Discovery
| Tool | Description |
|---|---|
find |
Run a find query with filter, projection, sort, and limit |
aggregate |
Execute an aggregation pipeline |
count |
Count documents matching a filter |
collection-schema |
Inspect the inferred schema of a collection |
collection-indexes |
List all indexes on a collection |
list-databases |
List all databases |
list-collections |
List collections in a database |
db-stats |
Get database statistics |
collection-storage-size |
Get the size of a collection in MB |
explain |
Analyze query execution plan |
export |
Export query or aggregation results as EJSON |
mongodb-logs |
Return recent mongod log events |
Write Operations
| Tool | Description |
|---|---|
insert-many |
Insert an array of documents |
update-many |
Update all documents matching a filter |
delete-many |
Delete all documents matching a filter |
create-collection |
Create a new collection |
drop-collection |
Remove a collection and its indexes |
drop-database |
Remove a database |
rename-collection |
Rename a collection |
create-index |
Create a standard or vector search index |
drop-index |
Drop an index by name |
Connection Management
| Tool | Description |
|---|---|
connect |
Connect to a MongoDB instance or Atlas cluster |
switch-connection |
Switch to a different MongoDB connection |
Atlas Management
| Tool | Description |
|---|---|
atlas-list-orgs |
List Atlas organizations |
atlas-list-projects |
List Atlas projects |
atlas-create-project |
Create an Atlas project |
atlas-list-clusters |
List Atlas clusters |
atlas-inspect-cluster |
Inspect cluster metadata |
atlas-create-free-cluster |
Create a free-tier Atlas cluster |
atlas-list-db-users |
List database users |
atlas-create-db-user |
Create a database user |
atlas-inspect-access-list |
View IP access list |
atlas-create-access-list |
Add IPs/CIDR ranges to access list |
atlas-list-alerts |
List active Atlas alerts |
atlas-get-performance-advisor |
Get index recommendations from Performance Advisor |
Atlas Local Development
| Tool | Description |
|---|---|
atlas-local-create-deployment |
Spin up a local Atlas deployment via Docker |
atlas-local-list-deployments |
List local deployments |
atlas-local-connect-deployment |
Connect to a local deployment |
atlas-local-delete-deployment |
Delete a local deployment |
The power ships with skills adapted from mongodb/agent-skills — the canonical best-practice guidance used across MongoDB agent integrations. Reference a skill by name in Kiro chat to activate it (e.g. #mongodb-schema-design).
Schema Design
- Choose between embedding and referencing based on access patterns and document growth
- Apply proven patterns: bucket, computed, polymorphic, outlier, archive, extended reference
- Detect anti-patterns: excessive lookups, unnecessary collections, redundant indexes
- Enforce structure with JSON Schema validation
Query Optimization
- Diagnose slow queries using
explainplans and Atlas Performance Advisor - Build compound indexes following the ESR (Equality → Sort → Range) rule
- Eliminate COLLSCAN plans and identify unused indexes
Natural Language Querying
- Translate plain-English questions into schema-validated
findqueries and aggregation pipelines - Index-aware optimization: suggests filters that align with existing indexes
Connection Configuration
- Tune connection pool sizes, timeout values, and TLS/SSL settings
- Profiles for serverless, OLTP, OLAP, and bursty workloads
- Covers all supported driver languages
MCP Server Setup
- Configure the MongoDB MCP Server with connection strings or Atlas Service Account credentials
- Set up Atlas Local deployments for development
- Enable read-only mode for safer access
Atlas Stream Processing
- Build real-time pipelines connecting Kafka, S3, Kinesis, and Lambda to MongoDB
- Provision Atlas Stream Processing workspaces and choose the right tier
- Debug processors and optimize pipeline performance
Search & AI
- Create Atlas Search indexes for full-text, autocomplete, and faceted search
- Build
$vectorSearchpipelines for semantic similarity and RAG applications - Implement hybrid search combining lexical and vector scoring
.
├── POWER.md # Power manifest — capabilities, tools, workflows, config
├── mcp.json # MCP server configuration
├── LICENSE # Apache 2.0
├── steering/ # Skills (loaded on demand in Kiro)
│ ├── atlas-stream-processing/
│ │ ├── skill.md # Stream processing skill
│ │ └── references/ # Connection configs, pipeline patterns, sizing guides
│ ├── mongodb-connection/
│ │ ├── skill.md # Connection optimization skill
│ │ └── references/ # Monitoring guide
│ ├── mongodb-mcp-setup/
│ │ └── skill.md # MCP server setup skill
│ ├── mongodb-natural-language-querying/
│ │ └── skill.md # NLQ skill
│ ├── mongodb-query-optimizer/
│ │ ├── skill.md # Query optimization skill
│ │ └── references/ # Indexing principles, aggregation optimization, anti-patterns
│ ├── mongodb-schema-design/
│ │ ├── skill.md # Schema design skill
│ │ └── references/ # 18 pattern, fundamental, and anti-pattern reference files
│ └── mongodb-search-and-ai/
│ ├── skill.md # Search & AI skill
│ └── references/ # Vector search, lexical search, hybrid search guides
├── tests/ # Property-based and integration tests
│ ├── skills.test.js # Validates skill file structure and POWER.md integrity
│ ├── skill-invocation-syntax.pbt.test.js
│ ├── *-eval-data.js # Evaluation datasets per skill
│ ├── *-eval-coverage.test.js # Coverage tests per skill
│ └── *-validation-report.md # Validation reports
└── .kiro/ # Kiro workspace config (hooks, specs, settings)
| Issue | Solution |
|---|---|
| Connection refused | Verify the connection string and that the MongoDB instance is reachable |
| Authentication failed | Check username/password in the connection string or Atlas API credentials |
| Atlas API 403 | Ensure the Service Account has the required project/org role in Atlas |
Namespace not found |
Use list-databases and list-collections to verify exact names |
| Index build failed | Check for duplicate key violations with find; verify field paths |
| Slow aggregation | Add an early $match stage to reduce documents processed; add indexes |
COLLSCAN on explain |
Create a compound index aligned to your filter and sort fields using the ESR rule |
| Docker not running | Start Docker Desktop before using atlas-local-create-deployment |
| MCP server not connecting | Reconnect from Kiro's MCP panel; verify environment variables are set correctly |
- MongoDB MCP Server Documentation
- MongoDB MCP Server GitHub
- MongoDB Atlas Documentation
- MongoDB Aggregation Pipeline Reference
- mongodb/agent-skills
- Kiro IDE
- License: Apache-2.0
- MCP Server License: Apache-2.0
- Privacy Policy: MongoDB Privacy Policy
- Support: GitHub Issues | MongoDB Community Forums