# Youtube Title Tag Optimizer > Optimize YouTube video titles, tags, and descriptions before publishing using YouTube Data API v3. Analyze top-ranking videos for a keyword to reverse-engineer winning title patterns, extract effective tags, and generate optimized title variations. Use when users want to (1) Optimize a video title before publishing, (2) Find the best tags for a video, (3) Analyze what title patterns work for a ke… - **Type:** Skill - **Install:** `agentstack add skill-nikhilbhansali-youtube-data-skills-youtube-title-tag-optimizer` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [nikhilbhansali](https://agentstack.voostack.com/s/nikhilbhansali) - **Installs:** 0 - **Category:** [Content & Media](https://agentstack.voostack.com/c/content-and-media) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [nikhilbhansali](https://github.com/nikhilbhansali) - **Source:** https://github.com/nikhilbhansali/youtube-data-skills/tree/master/skills/youtube-title-tag-optimizer ## Install ```sh agentstack add skill-nikhilbhansali-youtube-data-skills-youtube-title-tag-optimizer ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # YouTube Title & Tag Optimizer Optimize your video title, tags, and description before publishing by analyzing what works for top-ranking videos. ## Usage ``` /youtube-title-tag-optimizer "air fryer recipes" /youtube-title-tag-optimizer "Python tutorial" --my-title "Learn Python in 10 Minutes" /youtube-title-tag-optimizer --keyword "home workout" --my-title "Best Home Workout for Beginners 2024" ``` ## Instructions When the user invokes this skill: ### Step 1: Parse Arguments Extract from the user's input: - **Keyword** (required): The topic/keyword to optimize for - **--my-title "..."** (optional): User's working title to score and improve ### Step 2: Get API Key Check the user's Claude memory for a YouTube Data API v3 key. If not found, ask: > "I need a YouTube Data API v3 key to analyze ranking videos. You can get one from the Google Cloud Console. Please paste your key." ### Step 3: Write the Script Write the following Python script to `/tmp/_yt_title_optimizer_XXXX.py` (where XXXX is a random suffix, e.g. `$(openssl rand -hex 4)`): ```python #!/usr/bin/env python3 """ YouTube Title & Tag Optimizer Analyzes top-ranking videos for a keyword to extract winning title patterns, effective tags, and generate optimization recommendations. Usage: YT_API_KEY=KEY python3 /tmp/_yt_title_optimizer_XXXX.py "keyword" YT_API_KEY=KEY python3 /tmp/_yt_title_optimizer_XXXX.py "keyword" --my-title "My Title" """ import argparse import json import os import re import sys from datetime import datetime, timezone from collections import Counter try: from googleapiclient.discovery import build from googleapiclient.errors import HttpError except ImportError: print("ERROR: google-api-python-client not installed. Run: pip3 install google-api-python-client") sys.exit(1) def parse_duration(iso_duration: str) -> int: m = re.match(r"PT(?:(\d+)H)?(?:(\d+)M)?(?:(\d+)S)?", iso_duration or "") if not m: return 0 return int(m.group(1) or 0) * 3600 + int(m.group(2) or 0) * 60 + int(m.group(3) or 0) def format_number(n: int) -> str: if n >= 1_000_000: return f"{n/1_000_000:.1f}M" if n >= 1_000: return f"{n/1_000:.1f}K" return str(n) def analyze_title(title: str) -> dict: """Analyze a single title for patterns.""" analysis = { "length_chars": len(title), "length_words": len(title.split()), "has_number": bool(re.search(r'\d', title)), "has_question": title.rstrip().endswith("?"), "has_how_to": bool(re.search(r'how\s+to', title, re.I)), "has_brackets": bool(re.search(r'[\[\(]', title)), "has_parentheses": bool(re.search(r'\(', title)), "has_caps_word": bool(re.search(r'\b[A-Z]{2,}\b', title)), "has_emoji": bool(re.search(r'[^\w\s,.\-!?\'\"()\[\]:;/\\@#$%^&*+=~`|<>]', title)), "has_year": bool(re.search(r'20[12]\d', title)), "has_list_format": bool(re.search(r'^\d+\s', title)), "has_vs": bool(re.search(r'\bvs\.?\b', title, re.I)), "has_colon": ":" in title, "has_pipe": "|" in title, "has_dash_separator": " - " in title or " — " in title, "has_exclamation": "!" in title, "starts_with_how": title.lower().startswith("how"), "starts_with_why": title.lower().startswith("why"), "starts_with_what": title.lower().startswith("what"), "starts_with_number": bool(re.match(r'^\d', title)), } # Detect power words power_words = { "urgency": ["now", "today", "immediately", "urgent", "hurry", "fast", "quick", "instantly", "stop"], "curiosity": ["secret", "hidden", "surprising", "shocking", "unexpected", "weird", "strange", "mystery", "truth", "reveal"], "value": ["free", "best", "top", "ultimate", "complete", "guide", "hack", "tips", "tricks", "easy", "simple"], "emotional": ["amazing", "incredible", "insane", "mind-blowing", "life-changing", "game-changer", "beautiful", "perfect", "love", "hate"], "fear": ["mistake", "wrong", "avoid", "never", "worst", "dangerous", "warning", "risk", "fail", "scam", "don't"], } title_lower = title.lower() found_power_words = {} for category, words in power_words.items(): matches = [w for w in words if w in title_lower] if matches: found_power_words[category] = matches analysis["power_words"] = found_power_words # Detect title structure/formula structures = [] if re.match(r'^\d+\s', title): structures.append("listicle") if re.search(r'how\s+to', title, re.I): structures.append("how-to") if title.rstrip().endswith("?"): structures.append("question") if re.search(r'\bvs\.?\b', title, re.I): structures.append("comparison") if re.search(r'\breview\b', title, re.I): structures.append("review") if re.search(r'in\s+\d+\s+(minute|min|second|sec|hour|day|step)', title, re.I): structures.append("time-bound") if re.search(r'for\s+(beginners|newbies|noobs|starters)', title, re.I): structures.append("beginner-targeted") if re.search(r'(complete|ultimate|full|definitive)\s+guide', title, re.I): structures.append("comprehensive-guide") if re.search(r'\b(do|don\'t|never|stop|avoid)\b', title, re.I): structures.append("imperative") analysis["structures"] = structures return analysis def main(): parser = argparse.ArgumentParser(description="Optimize YouTube titles and tags") parser.add_argument("keyword", help="Keyword to optimize for") parser.add_argument("--my-title", default=None, help="Your working title to score") parser.add_argument("--output-dir", default=None, help="Output directory") args = parser.parse_args() api_key = os.environ.get("YT_API_KEY") if not api_key: print("ERROR: YT_API_KEY environment variable not set.") sys.exit(1) youtube = build("youtube", "v3", developerKey=api_key) quota_used = 0 print(f"Analyzing keyword: {args.keyword}") if args.my_title: print(f"Your title: {args.my_title}\n") # --- Search by relevance --- print("Fetching top-ranking videos (by relevance)...") try: relevance_resp = youtube.search().list( part="snippet", q=args.keyword, type="video", order="relevance", maxResults=50, ).execute() relevance_ids = [item["id"]["videoId"] for item in relevance_resp.get("items", [])] quota_used += 100 except HttpError as e: print(f"Search error: {e}") sys.exit(1) # --- Search by view count --- print("Fetching most-viewed videos...") try: viewcount_resp = youtube.search().list( part="snippet", q=args.keyword, type="video", order="viewCount", maxResults=50, ).execute() viewcount_ids = [item["id"]["videoId"] for item in viewcount_resp.get("items", [])] quota_used += 100 except HttpError as e: viewcount_ids = [] print(f"View count search error: {e}") # Combine and deduplicate all_ids = list(dict.fromkeys(relevance_ids + viewcount_ids)) print(f"Total unique videos: {len(all_ids)}") # --- Fetch video details --- print("Fetching video details...") videos = [] for i in range(0, len(all_ids), 50): batch = all_ids[i:i+50] try: resp = youtube.videos().list( part="snippet,statistics,contentDetails", id=",".join(batch), ).execute() videos.extend(resp.get("items", [])) quota_used += 1 except HttpError as e: print(f"Error: {e}") print(f" Got details for {len(videos)} videos") # --- Process videos --- processed = [] all_tags = Counter() title_analyses = [] for v in videos: views = int(v.get("statistics", {}).get("viewCount", 0)) likes = int(v.get("statistics", {}).get("likeCount", 0)) comments = int(v.get("statistics", {}).get("commentCount", 0)) duration = parse_duration(v.get("contentDetails", {}).get("duration", "")) title = v["snippet"]["title"] tags = v.get("snippet", {}).get("tags", []) for tag in tags: all_tags[tag.lower()] += 1 t_analysis = analyze_title(title) title_analyses.append(t_analysis) processed.append({ "video_id": v["id"], "title": title, "title_analysis": t_analysis, "channel": v["snippet"]["channelTitle"], "views": views, "likes": likes, "comments": comments, "engagement_rate": round((likes + comments) / max(views, 1) * 100, 2), "duration_sec": duration, "tags": tags, "tag_count": len(tags), "description_first_line": v["snippet"].get("description", "").split("\n")[0][:200], "published_at": v["snippet"]["publishedAt"], }) processed.sort(key=lambda x: x["views"], reverse=True) # --- Aggregate title analysis --- total = len(title_analyses) or 1 aggregate_title = { "avg_length_chars": round(sum(t["length_chars"] for t in title_analyses) / total, 1), "avg_length_words": round(sum(t["length_words"] for t in title_analyses) / total, 1), "pct_has_number": round(sum(1 for t in title_analyses if t["has_number"]) / total * 100, 1), "pct_question": round(sum(1 for t in title_analyses if t["has_question"]) / total * 100, 1), "pct_how_to": round(sum(1 for t in title_analyses if t["has_how_to"]) / total * 100, 1), "pct_brackets": round(sum(1 for t in title_analyses if t["has_brackets"]) / total * 100, 1), "pct_caps_word": round(sum(1 for t in title_analyses if t["has_caps_word"]) / total * 100, 1), "pct_emoji": round(sum(1 for t in title_analyses if t["has_emoji"]) / total * 100, 1), "pct_year": round(sum(1 for t in title_analyses if t["has_year"]) / total * 100, 1), "pct_list_format": round(sum(1 for t in title_analyses if t["has_list_format"]) / total * 100, 1), "pct_colon": round(sum(1 for t in title_analyses if t["has_colon"]) / total * 100, 1), "pct_pipe": round(sum(1 for t in title_analyses if t["has_pipe"]) / total * 100, 1), "pct_exclamation": round(sum(1 for t in title_analyses if t["has_exclamation"]) / total * 100, 1), } structure_counter = Counter() for t in title_analyses: for s in t["structures"]: structure_counter[s] += 1 aggregate_title["structures"] = {k: round(v/total*100, 1) for k, v in structure_counter.most_common()} power_category_counter = Counter() for t in title_analyses: for cat in t["power_words"]: power_category_counter[cat] += 1 aggregate_title["power_word_categories"] = {k: round(v/total*100, 1) for k, v in power_category_counter.most_common()} word_counter = Counter() stop_words = {"the", "a", "an", "is", "it", "in", "on", "at", "to", "for", "of", "and", "or", "but", "with", "you", "your", "my", "this", "that", "i", "me", "we", "how", "what", "why", "do", "does", "can", "will", "be", "are", "was", "not", "no", "so", "if"} for v in processed: words = re.findall(r'[a-zA-Z]{3,}', v["title"].lower()) for w in words: if w not in stop_words: word_counter[w] += 1 aggregate_title["common_title_words"] = word_counter.most_common(30) # --- Score user's title if provided --- my_title_score = None if args.my_title: my_analysis = analyze_title(args.my_title) score = 0 feedback = [] if 40 80: feedback.append("Title may be too long -- could get truncated in search results") else: score += 10 feedback.append("Title length is acceptable but could be optimized") keyword_lower = args.keyword.lower() title_lower = args.my_title.lower() if keyword_lower in title_lower: score += 20 if title_lower.startswith(keyword_lower) or title_lower[:20].find(keyword_lower) >= 0: score += 5 feedback.append("Keyword appears early in title -- great for SEO") else: feedback.append("Keyword is present in title") else: kw_words = keyword_lower.split() matches = sum(1 for w in kw_words if w in title_lower) if matches > 0: score += 10 feedback.append(f"Partial keyword match ({matches}/{len(kw_words)} words)") else: feedback.append("WARNING: Keyword not found in title -- critical for SEO") if my_analysis["has_number"]: score += 10 feedback.append("Contains a number -- increases click-through rate") if my_analysis["power_words"]: score += 10 cats = list(my_analysis["power_words"].keys()) feedback.append(f"Uses power words ({', '.join(cats)})") else: feedback.append("Consider adding power words for emotional impact") if my_analysis["structures"]: score += 10 feedback.append(f"Uses proven structure: {', '.join(my_analysis['structures'])}") else: feedback.append("Consider using a proven structure (how-to, listicle, question)") if my_analysis["has_brackets"]: score += 5 feedback.append("Uses brackets -- can boost CTR by 33%") if my_analysis["has_caps_word"]: score += 5 feedback.append("Strategic use of CAPS for emphasis") top_10_titles = [v["title"] for v in processed[:10]] common_patterns_in_top = set() for t in top_10_titles: ta = analyze_title(t) for s in ta["structures"]: common_patterns_in_top.add(s) matching_patterns = set(my_analysis["structures"]) & common_patterns_in_top if matching_patterns: score += 10 feedback.append(f"Matches top-performer patterns: {', '.join(matching_patterns)}") my_title_score = { "title": args.my_title, "score": min(score, 100), "analysis": my_analysis, "feedback": feedback, } # --- Build output --- output = { "keyword": args.keyword, "analyzed_at": datetime.now(timezone.utc).isoformat(), "total_videos_analyzed": len(processed), "aggregate_title_analysis": aggregate_title, "top_tags": all_tags.most_common(50), "tag_stats": { "unique_tags": len(all_tags), "avg_tags_per_video": round(sum(v["tag_count"] for v in processed) / len(processed), 1) if processed else 0, }, "top_videos": [ { "title": v["title"], "video_id": v["video_id"], "views": v["views"], "engagement_rate": v["engagement_rate"], "channel": v["channel"], "tag_count": v["tag_count"], "title_analysis": v["title_analysis"], "description_first_line": v["description_first_line"], } for v in processed[:20] ], "all_videos": processed, "my_title_score": my_title_score, "quota_used": { "total_estimated": quota_used, }, } # --- Save --- safe_kw = re.sub(r'[^a-zA-Z0-9_-]', '_', args.keyword)[:100] date_str = datetime.now().strftime("%Y%m%d") output_dir = args.output_dir or f"yt_optimize_{safe_kw}_{date_str}" os.makedirs(output_dir, exist_ok=True) output_file = os.path.join(output_dir, "title_tag_data.json") try: with open(output_file, "w", encoding="utf-8") as f: json.dump(output, f, indent=2, ensure_ascii=False) except (IOErr … ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [nikhilbhansali](https://github.com/nikhilbhansali) - **Source:** [nikhilbhansali/youtube-data-skills](https://github.com/nikhilbhansali/youtube-data-skills) - **License:** MIT Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** yes - **Shell / process execution:** no - **Environment & secrets:** yes - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-nikhilbhansali-youtube-data-skills-youtube-title-tag-optimizer - Seller: https://agentstack.voostack.com/s/nikhilbhansali - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.