mirror of
https://github.com/hanzoai/skills.git
synced 2026-11-19 00:47:36 +00:00
Four MIT collections ported — growthOS, openclaw-marketing-skills, Marketing-Skills and pm-skills — as 160 skills, 12 agents and 45 commands, each naming its origin in its own frontmatter so the tree can be checked against THIRD_PARTY_NOTICES rather than trusted. Sixteen names appeared in more than one collection and none of the copies were identical; the longest wins, which lands on the more developed text where they genuinely diverge. A fifth collection is deliberately absent. awesome-growth-hacking-skills carries NO licence, which leaves its author every right by default — and it is a list of links rather than a collection of skills, so there is nothing in it to port and no permission to port it with. It is cited as a directory to read. THE CONNECTOR SKILLS ARE REWRITTEN, not ported. They arrived installing vendor SDKs, running OAuth flows of their own, and ending with long-lived credentials in JSON files under .agents/ and ~/.growthos/ at mode 600. That is a secret on a disk with no rotation path, protected from other users and not from the agent — and one auth implementation per skill. Hanzo holds these connectors in Go, per user, sealed in KMS, so `connect` is the one home for the flow and the platform skills defer to it by provider id: google_ads, meta_ads, x. It is reachable as HTTP or as MCP tools at api.hanzo.ai/v1/mcp — the same routes, typed once in Go, so a tool and a curl cannot disagree. Two had nothing real behind them and now say so. There is no search_console provider, so that skill names what does answer the same questions (google_analytics, google_bigquery) and states that the remedy is a provider in hanzoai/cloud rather than a vendor SDK here. And getxapi-connect was a wrapper around a paid reseller of X's public data with the vendor's key in an env var; it is `listen` now, on our own x connector — a skill named after a vendor makes the vendor the capability and dies with it. The skills referenced paths that only existed inside the source repo, so the shared material they read is ported to reference/ and the paths rewritten. Three skills need the growthOS APPLICATION rather than its data; per the estate's own split an application belongs in hanzoai/ads, social, email or sms as a cloud plugin, so those three say what they need instead of carrying an import that resolves to nothing. Third-party product names are out of our prose and kept in the notices, which is where the licence requires them and the only place they belong. A skill's persona names a role, not somebody else's product. Co-authored-by: Hanzo Dev <[email protected]>
15 KiB
15 KiB
name, description, source
| name | description | source |
|---|---|---|
| platform-mastery | Platform-by-platform algorithm knowledge, best practices, and content adaptation rules for YouTube, LinkedIn, Twitter/X, Reddit, Instagram, GitHub, Threads, and StackOverflow. Use when optimizing content for a specific platform or adapting cross-platform content. | melgarafael/growthOS |
Platform Mastery Skill
Deep algorithm knowledge and content optimization rules for each social platform. Ensures content is native to each platform rather than cross-posted generically.
Platform Quick Reference
| Platform | Algorithm Priority | Content Sweet Spot | Tone |
|---|---|---|---|
| YouTube | Watch time + CTR | 8-15 min long-form | Educational, personality-driven |
| Comments > reactions | 1200-1500 chars + image | Professional, insight-driven | |
| Twitter/X | Engagement rate | 200-280 chars + media | Concise, opinionated |
| Upvote ratio + authenticity | Text posts, genuine value | Authentic, humble, no marketing | |
| Saves + shares | Carousel (10 slides) | Visual-first, aspirational | |
| GitHub | Stars + community engagement | README quality + releases | Technical, precise |
| Threads | Conversation + reposts | 300-500 chars | Conversational, casual |
| StackOverflow | Answer quality score | Canonical answers | Technical, helpful, factual |
YouTube
Algorithm Signals (Priority Order)
- Click-through rate (CTR): Thumbnail + title combination. Target >5% CTR.
- Average view duration: Percentage of video watched. Target >50%.
- Session time: Does your video lead to more YouTube watching? Playlists help.
- Engagement rate: Likes, comments, shares — comments weighted highest.
- Upload consistency: Regular schedule signals reliability to algorithm.
Content Rules
- Hook in first 5 seconds — no intros, logos, or "hey guys welcome back"
- Retention triggers every 60-90 seconds (pattern interrupts, open loops)
- Chapters (3+ timestamps starting at 0:00) — improves search + UX
- End screens in last 20 seconds — link to related content
- Thumbnail: 3-5 words max, readable at 100px width, face + emotion
- Title: <60 chars, front-load keyword, curiosity gap
Optimal Format
youtube:
length: 8-15 minutes (sweet spot for ad revenue + algorithm)
title_length: "<60 chars"
description: "first 150 chars critical (shown in search)"
tags: 10-15
thumbnail: "1280x720, high contrast, face + text"
chapters: "minimum 3, start at 0:00"
end_screen: "last 20s, 1-2 elements"
posting_time: "weekdays 2-4 PM viewer timezone"
Anti-Patterns
- Asking to "smash that like button" before delivering value
- Thumbnails with >5 words of text
- First 30 seconds of logos, channel intros, sponsor reads
- Generic tags ("tutorial", "how to")
Algorithm Signals (Priority Order)
- Comments (especially long comments): Highest weight signal
- Dwell time: Time spent reading your post
- Reactions: Likes and other reactions (comments > reactions)
- Shares/reposts: Signal virality to algorithm
- Profile authority: SSI score, connection relevance
Content Rules
- First 3 lines are everything — hook before "...see more" truncation
- Line breaks after every 1-2 sentences (mobile readability)
- End with a question — drives comments (algorithm's top signal)
- Personal stories outperform corporate content 3-5x
- Images increase engagement 2x (single image > carousel on LinkedIn)
- Native documents (PDF carousels) get high dwell time
- No external links in post body — put links in first comment
Post Format
[Hook line — bold claim, counterintuitive take, or personal story opener]
[blank line — critical for mobile]
[2-3 short paragraphs, each 1-2 sentences]
[Key insight or framework — optionally use bullet points]
• Point 1
• Point 2
• Point 3
[Personal reflection or lesson learned]
[Question to drive comments]
#[hashtag1] #[hashtag2] #[hashtag3]
Optimal Format
linkedin:
length: 1200-1500 chars (sweet spot)
hashtags: 3-5 (more looks spammy)
images: 1 (native upload, not link preview)
links: "in first comment, never in post body"
posting_time: "Tue-Thu 8-10 AM viewer timezone"
format: "short paragraphs, line breaks between"
hook: "first 210 chars (before ...see more)"
Anti-Patterns
- Starting with "I'm excited to announce..."
- Wall of text without line breaks
- External links in post body (kills reach)
- Hashtag stuffing (>5 hashtags)
- Corporate jargon ("synergy", "leverage", "disrupt")
Twitter/X
Algorithm Signals (Priority Order)
- Engagement rate: Replies + retweets + likes relative to impressions
- Media attachments: Images/video boost distribution
- Recency: Tweets decay fast — timing matters
- Thread completion: Full thread reads signal quality
- Profile engagement history: Algorithm favors accounts you interact with
Content Rules
- One idea per tweet — don't cram multiple concepts
- Strong opening — first 7 words determine if people read
- Threads for long-form: Tweet 1 is the hook, each tweet is self-contained
- Media boosts reach: Image, screenshot, or short video
- Quote tweets with commentary outperform plain retweets
- Reply to your own tweet with additional context or links
Tweet Formats
# Single tweet (high-impact)
[Bold statement or hot take]
[Supporting evidence or example in 1 sentence]
[Optional: image/screenshot]
# Thread format
🧵 [Hook — why should I read this thread?]
1/ [First point — must stand alone as a tweet]
2/ [Second point — builds on first]
[...]
n/ [Conclusion + CTA — follow, bookmark, retweet]
Optimal Format
twitter:
tweet_length: 200-280 chars (near limit performs well)
thread_length: 5-12 tweets
media: "1 image or 1 short video per tweet"
hashtags: 0-2 (hashtags less important than on other platforms)
posting_time: "weekdays 9-11 AM or 1-3 PM"
reply_strategy: "reply to own tweet with link, add context"
quote_tweet: "add 1-2 sentences of commentary"
Anti-Patterns
- Hashtag stuffing (Twitter isn't Instagram)
- Threads that could be a single tweet
- Starting tweets with "1/" without a hook
- Posting only links without commentary
- Engagement bait ("Like if you agree")
Algorithm Signals (Priority Order)
- Upvote-to-downvote ratio in first hour: Critical window
- Comment volume and quality: Active discussion signals value
- Community fit: Content matching subreddit culture
- Account karma + age: Spam filtering mechanism
- Report ratio: Low reports = trusted content
Content Rules (CRITICAL: Reddit is Anti-Marketing)
- NO marketing speak — Reddit users detect and punish self-promotion
- Value-first, brand-never: Share knowledge, not products
- Read subreddit rules before posting — each sub has specific requirements
- Authentic tone: Write like a human, not a brand
- Text posts outperform links in most subreddits
- Respond to every comment on your posts (signals engagement)
- The 10:1 rule: 10 genuine community interactions per 1 self-promotional post
Post Format
# Title: [Specific, descriptive, no clickbait — subreddit convention]
[Context: Why you're sharing this. Personal experience preferred.]
[The actual value: step-by-step, findings, data, analysis]
[What you learned or what surprised you]
[Open question: "Has anyone else experienced this?" or "What would you do differently?"]
Optimal Format
reddit:
title: "descriptive, follows subreddit conventions"
length: "varies by subreddit — check top posts for norm"
tone: "authentic, casual, humble"
self_promotion: "max 10% of activity"
posting_time: "Sun-Mon morning (US timezones)"
images: "imgur links for image posts (subreddit dependent)"
flair: "always use if subreddit requires it"
crosspost: "use native crosspost, not duplicate posts"
Anti-Patterns
- Sounding like a press release
- Using marketing buzzwords ("game-changer", "revolutionary")
- Posting the same content to multiple subreddits without adaptation
- Not reading subreddit rules before posting
- Ignoring comments on your posts
- New account with only self-promotional posts
Algorithm Signals (Priority Order)
- Saves: Strongest signal — content worth returning to
- Shares (DMs): Content worth sharing privately
- Comments: Engagement depth
- Reach to non-followers: Explore page potential
- Reel completion rate: For video content
Content Rules
- Visual-first — text is secondary to visual impact
- Carousels get highest engagement (10 slides, save-worthy)
- Reels for reach, carousels for engagement, Stories for community
- Caption structure: Hook line → value → CTA → hashtags
- Hashtag strategy: 5-15 relevant hashtags, mix of sizes
- Stories: 3-7 per day, use polls/questions for engagement
- Alt text: Always add for accessibility + SEO
Optimal Format
instagram:
post_types:
carousel: "highest engagement, 10 slides, educational"
reel: "highest reach, 15-30s, trending audio"
story: "community building, 3-7/day, interactive stickers"
static: "brand aesthetic, quotes, announcements"
caption_length: "150-300 chars (before ...more)"
hashtags: 5-15 in caption or first comment
image_ratio: "1:1 (feed), 9:16 (reels/stories), 4:5 (portrait)"
posting_time: "weekdays 11 AM-1 PM or 7-9 PM"
alt_text: "always include, descriptive"
Anti-Patterns
- Text-heavy images (Instagram is visual-first)
- Using all 30 hashtags (looks spammy)
- Ignoring Stories and Reels (feed-only strategy is dead)
- Cross-posting Twitter screenshots
- No alt text on images
GitHub
Algorithm Signals (Visibility)
- Stars: Primary popularity metric
- Forks: Community adoption signal
- Contributor count: Active community
- README quality: First impression, determines stars
- Release frequency: Active maintenance signal
- Issue response time: Community health
Content Rules
- README is your landing page — treat it like a conversion page
- Structured README: Badge bar → hero → install → quickstart → docs link
- Release notes: Detailed, categorized (Added, Changed, Fixed, Removed)
- Discussions: Use for community engagement, not just Issues
- Contributing guide: Lower barrier to contribution
- Issue templates: Structured bug reports and feature requests
README Format
# Project Name
[One-line description]
[Badges: CI status, version, license, downloads]
## Quick Start
[3-5 lines to get running — must be copy-pasteable]
## Features
[Bullet list of key features]
## Installation
[Step-by-step, multiple methods if applicable]
## Documentation
[Link to full docs]
## Contributing
[Link to CONTRIBUTING.md]
## License
[License type]
Optimal Format
github:
readme: "hero + badges + quickstart + features + install"
releases: "semantic versioning, categorized changelog"
issues: "templates for bug reports, feature requests"
discussions: "enabled, categories for Q&A and ideas"
topics: "5-10 relevant topics/tags"
social_preview: "1280x640 image"
license: "always include"
Threads
Algorithm Signals
- Reposts: Highest distribution signal
- Replies: Conversation depth
- Likes: Engagement signal
- Cross-posted from Instagram: Slight boost for connected accounts
- Recency: Fast-moving feed
Content Rules
- Conversational tone — Threads is casual, not professional
- Shorter than Twitter — 300-500 chars sweet spot
- No hashtags (minimal hashtag culture)
- Cross-post from Instagram for connected reach
- Early-adopter advantage — algorithm still favoring content
Optimal Format
threads:
length: 300-500 chars
tone: "casual, conversational, authentic"
hashtags: 0-2 (platform culture is minimal hashtags)
images: "optional, less visual than Instagram"
posting_time: "mirrors Instagram timing"
cross_post: "from Instagram for additional reach"
StackOverflow
Algorithm Signals
- Answer score: Upvotes - downvotes on answers
- Accepted answer: Marked by question asker
- Answer age + activity: Canonical answers that stay updated
- User reputation: Higher rep = more privileges + trust
Content Rules
- Answer the question directly first — explanation after
- Code examples are mandatory for programming questions
- Cite documentation — link to official sources
- Edit and improve — update answers as languages/frameworks evolve
- No opinions — factual, documented, reproducible answers
- Formatting matters — code blocks, headers, lists
Optimal Format
stackoverflow:
answer_structure: "direct answer → code → explanation → docs link"
tone: "technical, factual, helpful, no opinions"
code: "always include, tested, minimal reproduction"
links: "official docs preferred"
edits: "keep answers current as APIs change"
questions: "include version, error message, minimal reproduction"
Anti-Patterns
- Opinion-based answers without documentation backing
- Link-only answers
- "Did you try googling?" responses
- Answering without code examples
- Not including version numbers
Cross-Platform Adaptation Framework
When adapting content across platforms, transform — don't copy-paste:
adaptation:
source_platform: "[original platform]"
target_platform: "[target platform]"
transformations:
tone: "[adjust formality level]"
length: "[compress or expand]"
format: "[restructure for platform norms]"
visuals: "[add/remove visual elements]"
cta: "[platform-appropriate action]"
hashtags: "[platform-specific strategy]"
Cross-Platform Content Matrix
| Source | → Twitter/X | |||
|---|---|---|---|---|
| Blog post | Extract key insight + personal angle | Thread or key quote tweet | Summarize findings, ask for discussion | Carousel with key points |
| Video | Key takeaway + behind-scenes | Clip + hook | Text summary of findings | Reel (15-30s clip) |
| Data/research | Professional analysis post | Data point tweet + image | Deep analysis post | Infographic carousel |
| Product update | Feature benefit story | Quick announcement + screenshot | "We built X because Y" story | Before/after carousel |
Platform Algorithm Updates
Algorithm knowledge requires periodic updates. When referencing algorithm behavior:
- State the general principle (durable) separately from specific weights (temporal)
- Note that exact algorithm weights change frequently
- Focus on user behavior signals — these are more stable than algorithm internals
- Test assumptions with data rather than relying on hearsay