Mission Squad Docs

Appearance

Local Collections (MissionSquad Core) ​

Operate on embedded collections managed by MissionSquad Core (separate from Vector Stores). Collections are identified by safeName (the :collectionName param).

Endpoints ​

GET /v1/core/collections ​

Returns an array of EmbeddingCollection objects (API keys masked in objects).

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

POST /v1/core/collections/:collectionName/search ​

Body:

ts
{
  query: string,
  embeddingModelName: string,
  topK?: number
}

Returns:

ts
{
  success: boolean,
  results?: Array<{
    id?: string,
    text: string,
    _distance?: number,
    metadata?: Record<string, any>,
    fileName?: string,
    path?: string,
    lines?: [number, number],
    collection?: string,
    model?: string
  }>,
  message?: string
}

Example:

ts
await fetch("https://agents.missionsquad.ai/v1/core/collections/my-docs/search", {
  method: "POST",
  headers: { "x-api-key": process.env.MSQ_API_KEY!, "Content-Type": "application/json" },
  body: JSON.stringify({
    query: "vector database",
    embeddingModelName: "nomic-embed-text-v1.5",
    topK: 5
  })
});

Notes:

  • If embeddingModelName is unsupported, the server returns 400 with a clear message.

GET /v1/core/collections/:collectionName/diagnostics ​

Returns:

ts
{ success: true, diagnostics, collectionName, safeName }

POST /v1/core/collections/:collectionName/recover ​

Attempt recovery of a collection.

Body:

ts
{
  strategy?: "auto" | "repair" | "reembed",  // defaults to "auto"
  force?: boolean                             // defaults to false
}

Response includes flags such as strategyUsed, dataRecovered, and vectorCount when applicable.

See also ​