- PERSISTENCE
- MONITORING
- RECOVERY
- HUMAN
Short answer
How can my AI track workflow progress?
Your AI uses a dedicated container to write down where it is in a process and read back those updates later. This allows it to handle multi-step tasks without losing its place. You get a persistent record of every stage reached.
State management outcomes.
Where your data lands.
The AI manages the state within your specific container.
PERSISTENCE
Saved progress
The AI writes intermediate steps to the container. You won't lose context if a session resets.
MONITORING
Live status checks
Your agent queries the container to see which steps are finished. It knows exactly what to do next.
RECOVERY
Resuming tasks
If a process stops, the AI pulls the last known state. It picks up right where it left off.
HUMAN
Manual oversight
You can check the container yourself to verify the AI's progress. You decide when to override a state.
The workflow
What your AI does when a task starts.
The AI treats the container as its external memory for long-running operations.
Initialize state
The AI creates a new record to mark the beginning of a process.
create_documentUpdate progress
As each step finishes, the AI writes the new status to the container.
create_documentCheck status
The AI looks up existing records to see what has already been completed.
query_documentsRetrieve specific state
The AI pulls a single, specific progress record to verify a detail.
get_document
Try it
Copy these to start.
Use these prompts to begin managing your state.
Starting points
Swap the bracketed text for your actual container names or specific task IDs.
Azure Cosmos DB Container Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_sR5SFdydHWMVwBi2kmwRkK1Y9eFVwnJ0PxFJe5pj/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Azure Cosmos DB Container capabilities are ready to use.
{
"mcpServers": {
"azure-cosmos-db-container-mcp": {
"url": "https://edge.vinkius.com/vk_preview_sR5SFdydHWMVwBi2kmwRkK1Y9eFVwnJ0PxFJe5pj/mcp"
}
}
}Create a new workflow state in the 'orders' container for task ID 5501.
Check the 'deployment_logs' container to see if the status for 'web-server-01' is complete.
Get the current progress document for the 'user_onboarding' process.
Query the 'batch_jobs' container for all documents where status is 'pending'.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Azure Cosmos DB Container once, then ask.
Log in once to link your account. Your credentials stay encrypted, and you can start managing data immediately.
Connect Azure Cosmos DB Container to your AIFAQ
How the state tracking behaves
- 01
Can the AI delete my existing data?
No. The AI only has tools to create, get, and query documents. It cannot delete or modify existing records.
- 02
Does the AI manage my Azure subscription?
No. The AI only interacts with the data inside the containers you have already set up.
- 03
Can I see what the AI is doing in real time?
Yes. Since the AI writes to a standard container, you can view the documents directly in your Azure portal.
- 04
Can the AI create new containers for me?
No. You must create the containers in Azure first. The AI only works with the containers you provide.
- 05
How does the AI know which document to update?
The AI uses query tools to find the correct document ID before it performs further actions.
More questions about Azure Cosmos DB Container? See everything the Azure Cosmos DB Container Connector can do
Connect Azure Cosmos DB Container to Claude, Cursor, ChatGPT & more
