## TL;DR
Return a generator instead of a list: change `return [...]` to `yield` each Request, dict, or Item. The error fires because the callback returned a plain list (or a string/tuple), and Scrapy only accepts a single Request, BaseItem, dict, None, or an iterable of those.

```text
scrapy 'Spider must return Request, BaseItem, dict or None'
```

## Use this when
- The traceback shows TypeError: Spider must return Request, BaseItem, dict or None, got 'list'
- The got type is str, tuple, or generator-wrapper oddities after a refactor
- Items stop flowing after you changed a return to a list comprehension

## Not for this skill when
- The error is in a middleware or pipeline rather than a spider callback
- Requests fail at download time (DNS, 403, timeout) without this TypeError
- You are debugging why items are empty rather than why the callback crashes

## Steps
1. Find the callback named in the traceback and look for a bare `return` with more than one object. Expected: you find `return items_list`, `return (req, item)`, or `return response.text`.
2. Convert the return to yields: replace `return results` with a loop doing `yield x` for each element. Expected: the function becomes a generator and the TypeError disappears.
3. If you must build a list (e.g. for sorting), return the list only if every element is a Request, BaseItem, dict, or None. Expected: isinstance check on each element passes before the return.
4. Guard helpers: any helper called by the callback that builds Requests must also yield or return a list of valid types, never a tuple or string. Expected: no helper returns a bare string or tuple.
5. Re-run with `scrapy crawl myspider` and watch the log for scraped-item counts. Expected: item count increments; no TypeError.

### Alternate for step 2 (keep return, yield inside a helper)
If the callback is shared, make it call a small generator helper and `yield from helper(response)`. Expected: same fix, less restructuring.

## Variant phrasings
### Spider must return Request BaseItem dict or None got list
Convert the list return to per-element yields.
### scrapy parse callback TypeError got tuple
Tuples are not accepted either; unpack into yields or return a list of valid types.

## Tool/version compatibility
Applies to Scrapy 2.x (2.5 through 2.11+). The check lives in scrapy/utils/misc.py arg_to_iter via the engine's callback wrapper and is unchanged across these versions.

## Why it happens
Scrapy's engine wraps every callback result and validates each yielded item against the accepted types. A Python list is technically iterable, but the engine first checks the top-level return value: returning one list object (not yielding its elements) fails the type check. Returning a string is worse because strings are iterable character by character, so Scrapy rejects them outright with this error rather than yielding garbage.

## Edge cases / pitfalls
- `return None` implicitly (no return statement) is fine; `return` alone is fine.
- async def parse callbacks must still yield, not return lists; the same type check applies after awaiting.
- errback signatures take a Failure, not a Response; returning response-derived strings there triggers the same error.
- ItemLoader.load_item() returns a dict-like Item, which is accepted; a raw Selector is not.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_12K-tZTUU9faucVI-ZeoiQ