fix(step): use 'is not None' guard to preserve falsy return values in @step decorator

[devteamaegis]

Summary

Two fixes in the messaging layer:

1. step.py — falsy-output silently dropped (primary bug)

The @step decorator records the decorated function's return value via:

# Before
if result and not step.output:
    step.output = result

A plain truthiness guard (if result) silently drops any falsy-but-meaningful return value: 0, False, "", [], {}, etc.

Example:

@cl.step
async def count_errors(text: str) -> int:
    # returns 0 when text is clean — this output was NEVER persisted
    return len(find_errors(text))

The 0 return is falsy, so step.output is never set. The step appears to have no output in the UI, and the value is not persisted.

Fix: Use result is not None — only skip recording when the function explicitly returns nothing.

# After
if result is not None and not step.output:
    step.output = result

This applies to both the async_wrapper and sync_wrapper paths.

2. message.py — @classmethod names first parameter self

MessageBase.from_dict is decorated with @classmethod but its first parameter is named self instead of the conventional cls. While Python itself doesn't care about the parameter name, type-checkers (mypy, pyright) and linters flag this as an error, and it misleads readers into thinking self is an instance.

Renamed to cls following PEP 8 / Python convention.

---

Summary by cubic

Fixes the @step decorator to keep falsy but valid return values (e.g., 0, False, "", [], {}). Ensures step outputs are saved and shown; also aligns a classmethod signature with conventions to satisfy linters.

• Bug Fixes
  • backend/chainlit/step.py: use result is not None in both async and sync wrappers so falsy outputs are recorded.
  • backend/chainlit/message.py: rename self to cls in from_dict to follow classmethod convention and pass type-checkers.

^{Written for commit 98d81cb. Summary will update on new commits.}

[cubic-dev-ai]

No issues found across 2 files

_{Re-trigger cubic}