Mission Squad Docs

Appearance

Vector Stores ​

Vector stores let you create named collections of embedded files with configurable chunking strategies. Files are uploaded via the Files API and then associated with a vector store for embedding and search.

Endpoints ​

GET /v1/vector_stores ​

Returns a list of your vector stores.

ts
await fetch("https://agents.missionsquad.ai/v1/vector_stores", {
  headers: { "x-api-key": process.env.MSQ_API_KEY! }
});

POST /v1/vector_stores ​

Create a new vector store (if not existing), optionally enqueue files to embed.

Body:

ts
{
  name: string,
  file_ids?: string[],
  chunking_strategy?: { type: "auto" } | {
    type: "static",
    static: { max_chunk_size_tokens: number, chunk_overlap_tokens: number }
  },
  metadata?: Record<string, any>,
  embeddingModelName?: string,     // Optional embedding model to use when embedding files into this store
  enhancePDF?: boolean,            // Optional PDF preprocessing flag
  sseSessionId?: string,           // Optional; if provided, you can cancel via /v1/vector_stores/cancel
  batchSize?: number               // Optional; per-batch concurrency (server may have global limits)
}

Response: the VectorStore object (bytes/file_counts update as processing completes). If embeddingModelName is unsupported, the request returns 400 with an error message.

Example (Create store and add files):

ts
// 1) Upload files via /v1/files (see Files API)
// 2) Create a store with those file_ids
await fetch("https://agents.missionsquad.ai/v1/vector_stores", {
  method: "POST",
  headers: { "x-api-key": process.env.MSQ_API_KEY!, "Content-Type": "application/json" },
  body: JSON.stringify({
    name: "Research Papers",
    file_ids: ["file_abc123", "file_def456"],
    chunking_strategy: {
      type: "static",
      static: { max_chunk_size_tokens: 1024, chunk_overlap_tokens: 128 }
    },
    embeddingModelName: "nomic-embed-text-v1.5",
    enhancePDF: true,
    sseSessionId: "embed-session-1"
  })
});

GET /v1/vector_stores/:id ​

Return details for a vector store you own.

DELETE /v1/vector_stores/:id ​

Deletes the vector store, its core collection, associated file records, and attempts to remove on‑disk files.

GET /v1/vector_stores/:id/files ​

Lists files associated with the vector store.

POST /v1/vector_stores/:id/files ​

Add an existing uploaded file to the vector store and embed it.

Body:

ts
{
  file_id: string,
  chunking_strategy?: {
    type: "auto"
  } | {
    type: "static",
    static: { max_chunk_size_tokens: number, chunk_overlap_tokens: number }
  },
  enhancePDF?: boolean           // optional, defaults to the vector store setting if omitted
}

Notes:

  • Returns 409 if file is already in the store.

GET /v1/vector_stores/:id/files/:fileId ​

Returns the association record for a specific file in the store.

POST /v1/vector_stores/cancel ​

Cancel an in‑progress embedding session that was started with sseSessionId.

Body:

ts
{ sessionId: string }

Example:

ts
await fetch("https://agents.missionsquad.ai/v1/vector_stores/cancel", {
  method: "POST",
  headers: { "x-api-key": process.env.MSQ_API_KEY!, "Content-Type": "application/json" },
  body: JSON.stringify({ sessionId: "embed-session-1" })
});

See also ​