diff --git a/.claude/commands/pr-review.md b/.claude/commands/pr-review.md
new file mode 100644
index 00000000..b48fec4f
--- /dev/null
+++ b/.claude/commands/pr-review.md
@@ -0,0 +1,118 @@
+---
+allowed-tools: Read, Skill, Bash, Grep, Glob
+argument-hint: [extra_context]
+description: Review a GitHub pull request and display all issues found in Claude Code
+---
+
+# Review Pull Request
+
+Automatically review a GitHub pull request and display all found issues in Claude Code with detailed analysis and suggestions.
+
+## Usage
+
+```
+/pr-review [extra_context]
+```
+
+## Arguments
+
+- `extra_context` (optional): Additional instructions or filtering context (e.g., focus on specific issues or areas)
+
+## Examples
+
+```
+/pr-review                                    # Review all changes
+/pr-review Please focus on security issues    # Focus on security
+/pr-review Only review Python files           # Filter specific file types
+/pr-review Check for performance issues       # Focus on specific concern
+```
+
+## Instructions
+
+### 1. Auto-detect PR context
+
+- First check for environment variables:
+  - If `PR_NUMBER` and `GITHUB_REPOSITORY` are set, parse `GITHUB_REPOSITORY` as `owner/repo` and use `PR_NUMBER`
+  - Then use `gh pr view <PR_NUMBER> --repo <owner/repo> --json 'title,body'` to retrieve the PR title and description
+- Otherwise:
+  - Use `gh pr view --json 'title,body,url,number'` to get PR info for the current branch
+  - Parse the output to extract owner, repo, PR number, title, and description
+- If neither method works, inform the user that no PR was found and exit
+
+### 2. Fetch PR Diff and Existing Comments
+
+- Use `gh pr diff <PR_NUMBER> --repo <owner/repo>` to fetch the PR diff
+- Use the `fetch_unresolved_comments` skill to get existing unresolved review comments:
+  ```
+  /skill fetch_unresolved_comments <owner> <repo> <PR_NUMBER>
+  ```
+- Parse the returned comment data to identify lines that already have unresolved issues to avoid duplicate reports
+
+### 3. Review Changed Lines
+
+**Apply additional filtering** from user instructions if provided (e.g., focus on specific issues or areas)
+
+Carefully examine **only the changed lines** (added or modified) in the diff for:
+
+- Potential bugs and code quality issues
+- Common mistakes
+
+**Important**: Ignore unchanged/context lines and pre-existing code.
+
+**Collect all issues** in a structured format before presenting them:
+
+```
+issues = [
+  {
+    "file": "path/to/file.py",
+    "line": 42,
+    "end_line": 45, // (optional, for multi-line issues)
+    "severity": "error|warning|info", 
+    "category": "bug|style|performance|security",
+    "description": "Detailed issue description",
+    "suggestion": "Recommended fix",
+    "is_duplicate": false // true if similar issue already commented on this line
+  }
+]
+```
+
+**Important**: Before adding an issue to the list, check the existing unresolved comments from step 2. If there's already an unresolved comment on the same line(s) or covering a similar concern, mark the issue as duplicate to avoid redundant reports.
+
+### 4. Present Complete Analysis
+
+After reviewing all changed files, display a comprehensive summary in Claude Code:
+
+- **Group by file and severity** for clear organization
+- **Display statistics**: Total issues, breakdown by severity and category
+- **Use navigation-friendly format**: Include `file:line` references for easy IDE navigation
+- **Show detailed descriptions** and suggested fixes for each issue
+
+**Output format:**
+```
+## PR Review Results
+
+**Total Issues Found: X (Y new, Z already commented)**
+- 🔴 Errors: X
+- 🟡 Warnings: X  
+- 🔵 Info: X
+
+### path/to/file1.py:42-45
+🔴 **[Bug]:** Description here
+   → Suggestion: Fix recommendation
+
+### path/to/file2.js:67
+🟡 **[Style]:** Description here  
+   → Suggestion: Fix recommendation
+
+### path/to/file3.py:23
+🔴 **[Security]:** Description here
+   → Suggestion: Fix recommendation
+   ⚠️ *Similar issue already commented on this PR*
+```
+
+**Format Notes:**
+- Use `file:line` or `file:line-endline` format for direct IDE navigation
+- Mark duplicate/already-commented issues with ⚠️ warning
+- Group issues by file, with each issue as a separate section for easy clicking
+
+If **no issues found**, display: "✅ **No issues found** - All changes look good!"
diff --git a/.claude/skills/fetch-unresolved-comments/SKILL.md b/.claude/skills/fetch-unresolved-comments/SKILL.md
new file mode 100644
index 00000000..3700e9cc
--- /dev/null
+++ b/.claude/skills/fetch-unresolved-comments/SKILL.md
@@ -0,0 +1,88 @@
+---
+name: fetch-unresolved-comments
+description: Fetch unresolved PR review comments using GitHub GraphQL API, filtering out resolved and outdated feedback.
+---
+
+# Fetch Unresolved PR Review Comments
+
+Uses GitHub's GraphQL API to fetch only unresolved review thread comments from a pull request.
+
+## When to Use
+
+- You need to get only unresolved review comments from a PR
+- You want to filter out already-resolved and outdated feedback
+
+## Instructions
+
+1. **Parse PR information**:
+
+   - First check for environment variables:
+     - If `PR_NUMBER` and `GITHUB_REPOSITORY` are set, read them and parse `GITHUB_REPOSITORY` as `owner/repo` and use `PR_NUMBER` directly
+   - Otherwise:
+     - Use `gh pr view --json url -q '.url'` to get the current branch's PR URL and parse to extract owner, repo, and PR number
+
+2. **Run the Python script**:
+
+   ```bash
+   GITHUB_TOKEN=$(gh auth token) \
+       uv run python .claude/skills/fetch-unresolved-comments/fetch_unresolved_comments.py <owner> <repo> <pr_number>
+   ```
+
+3. **Script options**:
+
+   - `--token <token>`: Provide token explicitly (default: GITHUB_TOKEN or GH_TOKEN env var)
+
+4. **Parse the JSON output**:
+   The script always outputs JSON with:
+   - `total`: Total number of unresolved comments across all threads
+   - `by_file`: Review threads grouped by file path (each thread contains multiple comments in a conversation)
+
+## Example JSON Output
+
+```json
+{
+  "total": 3,
+  "by_file": {
+    ".github/workflows/resolve.yml": [
+      {
+        "thread_id": "PRRT_kwDOAL...",
+        "isOutdated": false,
+        "line": 40,
+        "startLine": null,
+        "diffHunk": "@@ -0,0 +1,245 @@\n+name: resolve...",
+        "comments": [
+          {
+            "id": 2437935275,
+            "body": "We can remove this once we get the key.",
+            "author": "harupy",
+            "createdAt": "2025-10-17T00:53:20Z"
+          },
+          {
+            "id": 2437935276,
+            "body": "Good catch, I'll update it.",
+            "author": "contributor",
+            "createdAt": "2025-10-17T01:10:15Z"
+          }
+        ]
+      }
+    ],
+    ".gitignore": [
+      {
+        "thread_id": "PRRT_kwDOAL...",
+        "isOutdated": false,
+        "line": 133,
+        "startLine": null,
+        "diffHunk": "@@ -130,0 +133,2 @@\n+.claude/*",
+        "comments": [
+          {
+            "id": 2437935280,
+            "body": "Should we add this to .gitignore?",
+            "author": "reviewer",
+            "createdAt": "2025-10-17T01:15:42Z"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
diff --git a/.claude/skills/fetch-unresolved-comments/fetch_unresolved_comments.py b/.claude/skills/fetch-unresolved-comments/fetch_unresolved_comments.py
new file mode 100755
index 00000000..5da47d1d
--- /dev/null
+++ b/.claude/skills/fetch-unresolved-comments/fetch_unresolved_comments.py
@@ -0,0 +1,169 @@
+"""Fetch unresolved PR review comments using GitHub GraphQL API.
+
+Example usage:
+    GITHUB_TOKEN=$(gh auth token) uv run --no-project .claude/skills/fetch-unresolved-comments/fetch_unresolved_comments.py mlflow mlflow 18327
+"""  # noqa: E501
+# ruff: noqa: T201
+
+import argparse
+import json
+import os
+import sys
+from typing import Any
+from urllib.request import Request, urlopen
+
+
+def fetch_unresolved_comments(owner: str, repo: str, pr_number: int, token: str) -> dict[str, Any]:
+    """Fetch unresolved review threads from a PR using GraphQL."""
+
+    query = """
+    query($owner: String!, $repo: String!, $prNumber: Int!) {
+      repository(owner: $owner, name: $repo) {
+        pullRequest(number: $prNumber) {
+          reviewThreads(first: 100) {
+            nodes {
+              id
+              isResolved
+              isOutdated
+              comments(first: 100) {
+                nodes {
+                  id
+                  databaseId
+                  body
+                  path
+                  line
+                  startLine
+                  diffHunk
+                  author {
+                    login
+                  }
+                  createdAt
+                  updatedAt
+                }
+              }
+            }
+          }
+        }
+      }
+    }
+    """
+
+    variables = {
+        "owner": owner,
+        "repo": repo,
+        "prNumber": pr_number,
+    }
+
+    headers = {
+        "Authorization": f"Bearer {token}",
+        "Content-Type": "application/json",
+    }
+
+    data = json.dumps({"query": query, "variables": variables}).encode("utf-8")
+    request = Request("https://api.github.com/graphql", data=data, headers=headers)
+
+    try:
+        with urlopen(request) as response:
+            return json.loads(response.read().decode("utf-8"))
+    except Exception as e:
+        print(f"Error fetching data: {e}", file=sys.stderr)
+        sys.exit(1)
+
+
+def format_comments(data: dict[str, Any]) -> dict[str, Any]:
+    """Format unresolved comments for easier consumption."""
+
+    try:
+        threads = data["data"]["repository"]["pullRequest"]["reviewThreads"]["nodes"]
+    except (KeyError, TypeError):
+        print("Error: Invalid response structure", file=sys.stderr)
+        print(json.dumps(data, indent=2), file=sys.stderr)
+        sys.exit(1)
+
+    by_file = {}
+    total_comments = 0
+
+    for thread in threads:
+        if not thread["isResolved"]:
+            comments = []
+            path = None
+            line = None
+            start_line = None
+            diff_hunk = None
+
+            for comment in thread["comments"]["nodes"]:
+                if path is None:
+                    path = comment["path"]
+                    line = comment["line"]
+                    start_line = comment.get("startLine")
+                    diff_hunk = comment.get("diffHunk")
+
+                comments.append(
+                    {
+                        "id": comment["databaseId"],
+                        "body": comment["body"],
+                        "author": comment["author"]["login"] if comment["author"] else "unknown",
+                        "createdAt": comment["createdAt"],
+                    }
+                )
+                total_comments += 1
+
+            if path:
+                if path not in by_file:
+                    by_file[path] = []
+
+                by_file[path].append(
+                    {
+                        "thread_id": thread["id"],
+                        "isOutdated": thread["isOutdated"],
+                        "line": line,
+                        "startLine": start_line,
+                        "diffHunk": diff_hunk,
+                        "comments": comments,
+                    }
+                )
+
+    return {
+        "total": total_comments,
+        "by_file": by_file,
+    }
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Fetch unresolved PR review comments using GitHub GraphQL API"
+    )
+    parser.add_argument("owner", help="Repository owner")
+    parser.add_argument("repo", help="Repository name")
+    parser.add_argument("pr_number", type=int, help="Pull request number")
+    parser.add_argument(
+        "--token",
+        default=os.environ.get("GITHUB_TOKEN") or os.environ.get("GH_TOKEN"),
+        help="GitHub token (default: GITHUB_TOKEN or GH_TOKEN env var)",
+    )
+
+    args = parser.parse_args()
+
+    if not args.token:
+        print(
+            "Error: GitHub token required (use --token or set GITHUB_TOKEN/GH_TOKEN)",
+            file=sys.stderr,
+        )
+        sys.exit(1)
+
+    data = fetch_unresolved_comments(args.owner, args.repo, args.pr_number, args.token)
+    formatted = format_comments(data)
+    formatted["by_file"] = {
+        path: [thread for thread in threads if not thread["isOutdated"]]
+        for path, threads in formatted["by_file"].items()
+    }
+    formatted["by_file"] = {k: v for k, v in formatted["by_file"].items() if v}
+    formatted["total"] = sum(
+        len(thread["comments"]) for threads in formatted["by_file"].values() for thread in threads
+    )
+
+    print(json.dumps(formatted, indent=2))
+
+
+if __name__ == "__main__":
+    main()