> ## Documentation Index
> Fetch the complete documentation index at: https://manager.harmix.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Error codes

> All retrieve_memory error_code values with meanings and recommended actions.

`retrieve_memory` tool responses may include these `error_code` values in `structuredContent`. All use HTTP 200 with `isError: true`.

| error\_code            | Meaning                                                       | Recommended action                                                  |
| ---------------------- | ------------------------------------------------------------- | ------------------------------------------------------------------- |
| `memory_not_ready`     | Memory is still provisioning or processing                    | Retry after memory is ready (see Memory MCP readiness)              |
| `no_sources_connected` | No memory sources connected for this account                  | Connect sources in Integrations, then retry                         |
| `quota_exceeded`       | Rate limit exceeded for Memory MCP                            | Back off and retry later; reduce call frequency                     |
| `entitlement_required` | Account has no remaining credits or subscription access       | Upgrade or renew your PAM subscription, then retry                  |
| `retrieval_timeout`    | Retrieval exceeded the server time limit (120s)               | Retry with a narrower prompt or fewer parallel calls                |
| `memory_unavailable`   | Temporary issue blocked retrieval                             | Retry after a short delay; contact support if it persists           |
| `invalid_arguments`    | Missing or invalid tool arguments (empty or oversized prompt) | Fix arguments per the [parameter table](/docs/retrieve-memory) and retry |

## Readiness states

When memory is not ready, `structuredContent.readiness_state` may indicate:

| State                 | Description                           |
| --------------------- | ------------------------------------- |
| `not_provisioned`     | Memory workspace not yet created      |
| `waiting_for_sources` | No integrations connected             |
| `processing`          | Initial sync and analysis in progress |
| `syncing_to_vm`       | Knowledge syncing for retrieval       |
| `ready_minimal`       | Ready with limited context            |
| `ready_rich`          | Fully ready                           |
| `error`               | Processing error — check Integrations |

Check current readiness on [Setup](https://pam.harmix.ai/setup) or via `GET /v1/dev/readiness`.

## HTTP vs tool errors

Memory MCP returns tool-level errors inside a successful HTTP 200 response. Only transport-level failures (network, auth header missing, invalid key) return non-200 HTTP status codes.

<Note>
  `quota_exceeded` applies only to `retrieve_memory` via `tools/call`, not to handshake methods or Developer REST calls.
</Note>