Check the token starts with `apify_api_`. Apify API tokens carry a fixed prefix; a token pasted without it (or a truncated copy) fails auth with "User was not found or authentication token is not valid". Re-copy the full token from the Console including the prefix, update the config, restart.

```text
401: User was not found or authentication token is not valid
```

## The fix

1. Open the Apify Console token page and copy the token in full, starting with `apify_api_`.
   Expected: the copied string begins with the prefix.
2. Replace the token value in your MCP config env block.
   Expected: the config still parses as valid JSON.
3. Fully restart the MCP client.
   Expected: Actor calls authenticate; the 401 is gone.

## When this applies

- 401 on every call with a token you typed or copied by hand.
- The stored token looks shorter than the one in the Console.

## When it does NOT apply

- Full token with prefix still 401s: the token was revoked; issue a new one.
- record-or-token-not-found: visibility problem, different skill.

## Tool and version compatibility

- @apify/actors-mcp-server and direct Apify API calls; the prefix rule is API-wide.

## Variant phrasings

### Apify 401 with a token I just created
Re-copy it. Fresh tokens are usually broken by a partial copy, not by the platform.

### User was not found but the user exists
The message is about the token, not the account. Fix the token string.

## Why it happens

The API parses the token format before looking anything up; a string without the expected prefix never matches a credential and returns the generic not-found message. It reads like an account problem but is always the token string.

## Edge cases

- Some password managers strip the prefix on autofill; paste into a scratch buffer and eyeball the start first.
- Tokens shown once at creation: if you lost the full string, create a new token rather than guessing.
- Rotate tokens you pasted into shared configs; the old one stays valid until revoked.