Add AI EP review GitHub Action

.github/scripts/ep_review.py

@@ -0,0 +1,236 @@
+#!/usr/bin/env python3
+"""
+AI EP Review — thin wrapper around the ep-review skill.
+
+Reads SKILL.md as the system prompt, fetches the PR diff,
+calls Claude API for structured review, sanitizes output,
+posts the result as a PR comment, and applies the reviewed label.
+"""
+
+import json
+import os
+import re
+import subprocess
+import sys
+import tempfile
+
+from anthropic import AnthropicVertex
+
+REPO = os.environ["REPO"]
+PR_NUMBER = os.environ["PR_NUMBER"]
+SKILL_PATH = ".osac-workspace/skills/ep-review/SKILL.md"
+TEMPLATE_PATH = "guidelines/enhancement_template.md"
+REVIEWED_LABEL = "rfe-creator-auto-reviewed"
+GCP_PROJECT = os.environ["GCP_PROJECT"]
+GCP_REGION = os.environ["GCP_REGION"]
+MODEL = "claude-sonnet-4@20250514"
+BOT_LOGIN = "github-actions[bot]"
+
+REVIEW_TOOL = {
+    "name": "submit_review",
+    "description": "Submit a structured EP review with scores and findings.",
+    "input_schema": {
+        "type": "object",
+        "properties": {
+            "title": {"type": "string", "description": "EP title from the PR"},
+            "scores": {
+                "type": "object",
+                "properties": {
+                    "what": {"type": "integer", "minimum": 0, "maximum": 2,
+                             "description": "Problem clarity and template compliance (0-2)"},
+                    "why": {"type": "integer", "minimum": 0, "maximum": 2,
+                            "description": "Business justification and user stories (0-2)"},
+                    "how": {"type": "integer", "minimum": 0, "maximum": 2,
+                            "description": "Approach, architecture alignment, acceptance criteria (0-2)"},
+                    "task": {"type": "integer", "minimum": 0, "maximum": 2,
+                             "description": "Enhancement quality, cross-cutting concerns (0-2)"},
+                    "size": {"type": "integer", "minimum": 0, "maximum": 2,
+                             "description": "Scope appropriateness, completeness (0-2)"},
+                },
+                "required": ["what", "why", "how", "task", "size"]
+            },
+            "criterionNotes": {
+                "type": "object",
+                "properties": {
+                    "what": {"type": "string"},
+                    "why": {"type": "string"},
+                    "how": {"type": "string"},
+                    "task": {"type": "string"},
+                    "size": {"type": "string"},
+                },
+                "required": ["what", "why", "how", "task", "size"]
+            },
+            "findings": {
+                "type": "object",
+                "properties": {
+                    "critical": {"type": "array", "items": {"type": "string"}},
+                    "important": {"type": "array", "items": {"type": "string"}},
+                    "suggestions": {"type": "array", "items": {"type": "string"}},
+                },
+                "required": ["critical", "important", "suggestions"]
+            }
+        },
+        "required": ["title", "scores", "criterionNotes", "findings"]
+    }
+}
+
+
+def gh(args):
+    """Run a gh CLI command and return stdout."""
+    result = subprocess.run(
+        ["gh"] + args, capture_output=True, text=True, timeout=60
+    )
+    if result.returncode != 0:
+        print(f"gh command failed with exit code {result.returncode}", file=sys.stderr)
+        sys.exit(1)
+    return result.stdout
+
+
+def read_file(path):
+    """Read and return the contents of a file."""
+    with open(path) as f:
+        return f.read()
+
+
+def sanitize_text(text, max_len=500):
+    """Strip dangerous markdown constructs from model output."""
+    text = re.sub(r'!\[[^\]]*\]\([^\)]*\)', '', text)
+    text = re.sub(r'\[([^\]]*)\]\([^\)]*\)', r'\1', text)
+    text = re.sub(r'<[^>]+>', '', text)
+    text = re.sub(r'@(\w+)', r'\1', text)
+    text = re.sub(r'https?://(?!redhat\.atlassian\.net)\S+', '[link removed]', text)
+    return text.strip()[:max_len]
+
+
+def validate_review(review):
+    """Clamp scores to valid range and sanitize all string fields."""
+    for k in ["what", "why", "how", "task", "size"]:
+        review["scores"][k] = max(0, min(2, int(review["scores"][k])))
+    review["title"] = sanitize_text(review.get("title", ""), 200)
+    for k in review.get("criterionNotes", {}):
+        review["criterionNotes"][k] = sanitize_text(review["criterionNotes"][k], 200)
+    for sev in ["critical", "important", "suggestions"]:
+        review["findings"][sev] = [sanitize_text(f) for f in review["findings"].get(sev, [])]
+    return review
+
+
+def format_comment(review):
+    """Format a validated review into a markdown PR comment."""
+    title = review["title"]
+    s = review["scores"]
+    total = sum(s.values())
+    verdict = "PASS" if total >= 5 else "FAIL"
+    notes = review["criterionNotes"]
+    findings = review["findings"]
+
+    lines = [
+        f"## AI EP Review: {title}",
+        "",
+        f"**Score: {total}/10** | **Verdict: {verdict}**",
+        "",
+        "| Criterion | Score | Notes |",
+        "|-----------|-------|-------|",
+    ]
+    for key in ["what", "why", "how", "task", "size"]:
+        label = key.capitalize()
+        note = notes.get(key, "").replace("|", "\\|").replace("\n", " ")
+        lines.append(f"| {label} | {s[key]}/2 | {note} |")
+
+    for severity in ["critical", "important", "suggestions"]:
+        items = findings.get(severity, [])
+        lines.append("")
+        lines.append(f"### {severity.capitalize()} ({len(items)})")
+        if items:
+            for i, item in enumerate(items, 1):
+                lines.append(f"{i}. {item}")
+        else:
+            lines.append("None.")
+
+    return "\n".join(lines)
+
+
+def main():
+    """Run the EP review: read inputs, call Claude, sanitize, post comment, apply label."""
+    skill_prompt = read_file(SKILL_PATH)
+    template = read_file(TEMPLATE_PATH)
+
+    diff_file = os.environ.get("PR_DIFF_FILE")
+    meta_file = os.environ.get("PR_META_FILE")
+    if diff_file and meta_file:
+        diff = read_file(diff_file)
+        pr_info = json.loads(read_file(meta_file))
+    else:
+        diff = gh(["pr", "diff", PR_NUMBER, "--repo", REPO])
+        pr_meta = gh(["pr", "view", PR_NUMBER, "--repo", REPO,
+                      "--json", "title,body,author,files"])
+        pr_info = json.loads(pr_meta)
+
+    if not diff.strip():
+        print("No diff found — skipping review.")
+        return
+
+    truncated = len(diff) > 50000
+    if truncated:
+        print(f"Diff truncated from {len(diff)} to 50000 chars")
+
+    client = AnthropicVertex(project_id=GCP_PROJECT, region=GCP_REGION)
+    user_message = (
+        "Review this Enhancement Proposal PR against the OSAC EP template.\n\n"
+        f"## PR: {pr_info['title']}\n\n"
+        f"## EP Template\n\n{template}\n\n"
+        f"## PR Diff\n\n```\n{diff[:50000]}\n```\n\n"
+        "Use the submit_review tool to provide your structured review."
+    )
+
+    print(f"Reviewing EP: {pr_info['title']}")
+
+    response = client.messages.create(
+        model=MODEL,
+        max_tokens=4096,
+        system=skill_prompt,
+        tools=[REVIEW_TOOL],
+        messages=[{"role": "user", "content": user_message}],
+    )
+
+    review = None
+    for block in response.content:
+        if block.type == "tool_use" and block.name == "submit_review":
+            review = block.input
+            break
+
+    if not review:
+        print("No structured review returned — model did not call submit_review tool.", file=sys.stderr)
+        sys.exit(1)
+
+    review = validate_review(review)
+    comment = format_comment(review)
+    if truncated:
+        comment += "\n\n> **Note:** PR diff was truncated to 50,000 characters. Review may be incomplete."
+    total = sum(review["scores"].values())
+    print(f"Review complete: {total}/10 ({'PASS' if total >= 5 else 'FAIL'})")
+
+    existing = gh(["api", f"repos/{REPO}/issues/{PR_NUMBER}/comments",
+                   "--jq", f'[.[] | select(.user.login == "{BOT_LOGIN}") | select(.body | startswith("## AI EP Review:"))][0].id // empty'])
+    comment_id = existing.strip()
+
+    with tempfile.NamedTemporaryFile(mode="w", suffix=".md", delete=False) as f:
+        f.write(comment)
+        comment_file = f.name
+
+    if comment_id:
+        gh(["api", f"repos/{REPO}/issues/comments/{comment_id}",
+            "--method", "PATCH", "--field", f"body=@{comment_file}"])
+        print(f"Updated existing review comment on PR #{PR_NUMBER}")
+    else:
+        gh(["pr", "comment", PR_NUMBER, "--repo", REPO, "--body-file", comment_file])
+        print(f"Posted new review comment on PR #{PR_NUMBER}")
+
+    os.unlink(comment_file)
+
+    gh(["pr", "edit", PR_NUMBER, "--repo", REPO,
+        "--add-label", REVIEWED_LABEL])
+    print(f"Applied label: {REVIEWED_LABEL}")
+
+
+if __name__ == "__main__":
+    main()

.github/workflows/ep-review.yml

@@ -0,0 +1,65 @@
+name: EP Review
+
+on:
+  pull_request_target:
+    types: [opened, synchronize, reopened]
+    paths:
+      - 'enhancements/**/*.md'
+
+concurrency:
+  group: ep-review-${{ github.event.pull_request.number }}
+  cancel-in-progress: true
+
+jobs:
+  review:
+    runs-on: ubuntu-latest
+    if: >-
+      github.event.pull_request.author_association == 'MEMBER' ||
+      github.event.pull_request.author_association == 'COLLABORATOR' ||
+      github.event.pull_request.author_association == 'OWNER'
+    permissions:
+      contents: read
+      pull-requests: write
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          ref: ${{ github.event.pull_request.base.sha }}
+          persist-credentials: false
+
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          repository: osac-project/osac-workspace
+          path: .osac-workspace
+          persist-credentials: false
+          sparse-checkout: |
+            skills/ep-review/SKILL.md
+
+      - uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5.6.0
+        with:
+          python-version: "3.13"
+
+      - run: pip install --quiet 'anthropic[vertex]==0.52.0'
+
+      - uses: google-github-actions/auth@ba79af03b8d7f2e92b05e4dbc41e536528e46e57 # v2.1.10
+        with:
+          credentials_json: ${{ secrets.GCP_SA_KEY }}
+
+      - name: Fetch PR diff
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          PR_NUMBER: ${{ github.event.pull_request.number }}
+          REPO: ${{ github.repository }}
+        run: |
+          gh pr diff "$PR_NUMBER" --repo "$REPO" > /tmp/pr-diff.txt
+          gh pr view "$PR_NUMBER" --repo "$REPO" --json title,body,author,files > /tmp/pr-meta.json
+
+      - name: Run AI EP Review
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          PR_NUMBER: ${{ github.event.pull_request.number }}
+          REPO: ${{ github.repository }}
+          GCP_PROJECT: ${{ secrets.GCP_PROJECT }}
+          GCP_REGION: ${{ secrets.GCP_REGION }}
+          PR_DIFF_FILE: /tmp/pr-diff.txt
+          PR_META_FILE: /tmp/pr-meta.json
+        run: python .github/scripts/ep_review.py