feat: add script to generate PR summaries from git diffs

Reads a diff from stdin and uses the OpenAI Responses API to output a
structured markdown PR description using a fixed system prompt.

Author: jwood000

ai_git_helpers/pr_summary.py

@@ -0,0 +1,121 @@
+#!/usr/bin/env python3
+
+import os
+import sys
+from openai import OpenAI
+
+MODEL = "gpt-5.2"
+
+SYSTEM_PROMPT = """
+You are helping maintain the R package RcppAlgos. You are an expert C++
+developer, algorithm designer, and understand the R and CRAN ecosystem deeply.
+
+Your task is to generate a professional GitHub Pull Request summary from a git diff.
+
+CRITICAL REQUIREMENTS:
+
+• This summary will be used as the PR description.
+• Focus on WHAT changed and WHY.
+• Be accurate and specific.
+• Do NOT invent behavior not supported by the diff.
+• If intent is inferred, state it as inference.
+
+• Highlight important technical implications:
+  - correctness
+  - edge cases
+  - performance
+  - memory safety
+  - API or behavioral changes
+
+• Call out anything reviewers should pay special attention to.
+
+• Prefer clarity over verbosity.
+• Do NOT include code blocks.
+• Do NOT include raw diff content.
+• Do NOT mention file paths unless absolutely necessary for clarity.
+
+Use the following structure exactly:
+
+## Summary
+
+High-level explanation of the change and its purpose.
+
+## Key Changes
+
+• Bullet list of concrete technical changes
+
+## Motivation
+
+Identify the PRIMARY motivation that drove this PR.
+
+Pay special attention to new combinatorial algorithms and treat them as
+likely primary motivations.
+
+This should reflect the original idea or feature that initiated the work,
+not secondary refactors or improvements discovered along the way.
+
+Then include a subsection:
+
+### Secondary Improvements
+
+List improvements that were discovered opportunistically while implementing
+the primary change.
+
+Be careful to distinguish clearly between:
+
+• the original driving idea
+• incidental fixes, refactors, or optimizations
+
+If uncertain, infer carefully from the diff and state that it is inferred.
+
+## Impact
+
+Describe any impact on:
+
+• User-facing behavior
+• Performance
+• Correctness
+• Backwards compatibility
+
+If none, state "No user-visible impact expected."
+
+## Reviewer Notes
+
+Call out:
+
+• Risky areas
+• Edge cases
+• Logic changes
+• Performance-sensitive code
+• CRAN / R API considerations if applicable
+
+Be professional and concise.
+
+Output markdown only.
+"""
+
+def main():
+
+    diff = sys.stdin.read().strip()
+
+    if not diff:
+        print("No diff provided.", file=sys.stderr)
+        sys.exit(1)
+
+    api_key = os.getenv("RCPPALGOS_OPENAI_KEY")
+    client = OpenAI(api_key=api_key)
+
+    resp = client.responses.create(
+        model=MODEL,
+        input=[
+            {"role": "system", "content": SYSTEM_PROMPT},
+            {"role": "user", "content": diff},
+        ],
+        max_output_tokens=4000,
+    )
+
+    print(resp.output_text.strip())
+
+
+if __name__ == "__main__":
+    main()