Skip to content

Commit 32e7db2

Browse files
igerberclaude
andcommitted
Keep notebooks in AI review diff, exclude only data files
Notebooks contain methodology-relevant content that should be reviewed. Only benchmarks/data/real/*.json and *.csv need exclusion (golden values and survey data subsets that can exceed the 1MB input limit). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
1 parent 9a24a12 commit 32e7db2

1 file changed

Lines changed: 3 additions & 4 deletions

File tree

.github/workflows/ai_pr_review.yml

Lines changed: 3 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -148,12 +148,11 @@ jobs:
148148
git --no-pager diff --name-status "$BASE_SHA" "$HEAD_SHA"
149149
echo ""
150150
echo "Unified diff (context=5):"
151-
# Exclude large generated/data files from the full diff to stay
151+
# Exclude large generated data files from the full diff to stay
152152
# within the model's input limit. The --name-status above still
153-
# lists them. Narrowed to real-data assets and notebook outputs.
153+
# lists them so the reviewer knows they changed.
154154
git --no-pager diff --unified=5 "$BASE_SHA" "$HEAD_SHA" \
155-
-- . ':!benchmarks/data/real/*.json' ':!benchmarks/data/real/*.csv' \
156-
':!docs/tutorials/*.ipynb'
155+
-- . ':!benchmarks/data/real/*.json' ':!benchmarks/data/real/*.csv'
157156
} >> "$PROMPT"
158157
159158
- name: Run Codex

0 commit comments

Comments
 (0)