Skip to content

docs: clarify the wrap_llm README section - #26

Closed
tishachawla-jg wants to merge 1 commit into
mainfrom
docs/clarify-wrap-llm-main
Closed

docs: clarify the wrap_llm README section#26
tishachawla-jg wants to merge 1 commit into
mainfrom
docs/clarify-wrap-llm-main

Conversation

@tishachawla-jg

Copy link
Copy Markdown
Contributor

What

Fixes the wrap_llm section of the README, which was unclear on main.

The old text had a broken sentence ("...use wrap_llm gives the same record + on_crossing contract...") and never said plainly when you would reach for wrap_llm instead of a decorator.

After

The section now:

  • Leads with the one thing that decides it: use wrap_llm when your model call is a plain function you pass around (rather than a named function you can decorate with @boundary).
  • Spells out the two behaviors: Live runs and records an llm envelope; Replay (stubbed) returns the recorded result without calling the function.
  • Notes that messages / model / provider are read from the arguments automatically, so extract_input is only needed for unusual signatures.

Docs-only, single file. No em/en dashes.

The previous text had a broken sentence ("use wrap_llm gives the same
...") and did not make clear when to reach for wrap_llm. Rewrite it to
state plainly: use it when your model call is a plain function you pass
around, spell out live vs replay behavior, and note that inputs are read
by name automatically.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Tisha Chawla <tisha.chawla2020@vitalum.ac.in>
@tishachawla-jg

Copy link
Copy Markdown
Contributor Author

Superseded by #27, which restructures the whole README (including this wrap_llm section) and adds the I/O-vs-side-effects clarification. Closing in favor of that.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant