Repository navigation
Python: use AZURE_OPENAI_EMBEDDING_MODEL in the Foundry memory sample README - #9741
Open
NingLin Liu (ninglinLiu) wants to merge 1 commit into
Open
NingLin Liu (ninglinLiu) wants to merge 1 commit into
NingLin Liu (ninglinLiu) wants to merge 1 commit into
Conversation
NingLin Liu (ninglinLiu)
requested review from
SergeyMenshykh,
Tao Chen (TaoChenOSU),
Eduard van Valkenburg (eavanvalkenburg),
Giles Odigwe (giles17),
Jose Alvarez (jpalvarezl),
Evan Mattson (moonbox3),
Roger Barreto (rogerbarreto),
sophia-ramsey and
westey (westey-m)
as code owners
October 27, 2026 02:16
NingLin Liu (ninglinLiu)
deployed
to
github-app-auth
October 27, 2026 02:16 — with
GitHub Actions
Active
NingLin Liu (ninglinLiu)
deployed
to
github-app-auth
October 27, 2026 02:16 — with
GitHub Actions
Active
Contributor
There was a problem hiding this comment.
🟢 Approval recommended
The single-line documentation correction matches the sample’s environment-variable lookup and has no unresolved review issues.
No open issues 🎉
What changed in this PR
Corrects the Foundry memory sample’s prerequisites so the README matches the environment variable read by the sample.
Changes:
- Replaces
AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAMEwithAZURE_OPENAI_EMBEDDING_MODEL. - Documentation only; sample code and runtime behavior are unchanged.
| File | Description |
|---|---|
python/samples/02-agents/context_providers/README.md |
Corrects the embedding environment-variable name for the Foundry memory sample. |
Review effort: Balanced
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Motivation & Context
The context providers README lists
AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAMEas the embedding setting forazure_ai_foundry_memory.py, but the sample readsos.environ["AZURE_OPENAI_EMBEDDING_MODEL"]when it builds the memory store definition (and its own docstring says "Set AZURE_OPENAI_EMBEDDING_MODEL for the embedding model"). Following the README givesKeyError: 'AZURE_OPENAI_EMBEDDING_MODEL'before the memory store is created.Description & Review Guide
azure_ai_foundry_memory.pyprerequisites ofpython/samples/02-agents/context_providers/README.mdtoAZURE_OPENAI_EMBEDDING_MODEL.azure_ai_foundry_memory.pyline 58 (embedding_model=os.environ["AZURE_OPENAI_EMBEDDING_MODEL"]). The same name is also whatazure_ai_search/search_context_semantic.pyuses, so the README is now consistent across the context provider samples.Related Issue
N/A - trivial change. A stale environment-variable name in one README line; the correction follows the variable the existing sample already reads, with no behavior change.
AI Assistance
AI assistance details: finding the mismatch (README variables vs.
os.environreads in the samples), the documentation edit, and this PR description.Contribution Checklist
breaking changelabel (or add "[BREAKING]" to the title prefix, before or after any language prefix) — a workflow keeps the label and title prefix in sync automatically.