Meko for Aspire
Meko gives every agent in your Aspire app shared memory. Declare the datapack once in the AppHost. Every service that embeds an agent reads and writes the same store, in any language Aspire runs.
Over MCP, Meko provides Aspire apps with:
Agent-scoped memory
What your agent learns persists beyond the session and is readable via every AI tool you connect to.
A shared knowledge base
Documents and promoted conclusions are searchable by all agents and teammates with access to the datapack.
A trace of every read and write
Each conversation is stored with its reasoning, so you can see what an agent did and why.
The AppHost wires it, the datapack remembers it
Aspire models your distributed app in one file, and Meko does the same for what its agents learn.
A Python worker, a C# API service, and a TypeScript front end each run their own agent – and all three share one memory, knowledge base, and trace. The endpoint, API key, and datapack ID flow through Aspire’s own parameters. Operations covering memory, conversations, knowledge base, datapacks, artifacts, and observability are available to every service, and your code calls the same endpoint directly when an operation must happen every time.
How Meko Benefits Aspire
Give the whole app one memory
Each agent service connects to the same datapack, tagged by agent ID. What the ingest worker learns is available when the API agent answers.
Wire it once in the AppHost
Add the Meko endpoint and a secret API key parameter, then reference them from every agent service with WithEnvironment. Aspire prompts for missing values and saves them to user secrets.
Ship the same wiring everywhere
One AppHost definition runs locally and deploys. The Meko configuration travels with it.
Trace agents across service boundaries
Conversation tools record each turn with its reasoning, and artifact tools store the files a service produced next to it. Read both back from any tool on the datapack after the run.
Connecting Meko in Aspire
- Create a Meko API key and datapack in the Meko portal.
- In the AppHost, add a secret parameter for the key, and pass MEKO_API_KEY and MEKO_DATAPACK_ID to each agent service with WithEnvironment.
- In each agent service, connect its MCP client to https://mcp.mekodata.ai/mcp with an Authorization: Bearer <your-api-key> header and a User-Agent header (e.g. meko-aspire/1.0) – Meko’s edge rejects requests without one, and .NET’s HttpClient sends no User-Agent by default. Use the framework’s own clients: langchain-mcp-adapters or Pydantic AI in Python, the MCP client in .NET, etc.
- Run the app. Every agent service starts with the same memory.
Meko runs as a managed service, so your app model stays configuration-only. Aspire’s dashboard shows the injected environment variables per service, with secrets masked.