Agent skill
jab-hook
Quellcode ansehen: coreyhaines31/makerskills
Installation
npx skills add coreyhaines31/makerskills --skill jab-hook278
Installationen
EU-hosted inference API
Power your AI agent skills with open-source models.
Drop-in OpenAI-compatible API. No data leaves Europe.
MiniMax
MiniMax M3
$0.40 / $1.40
per M tokens
Z.ai
GLM 5.3 Flash
$0.20 / $0.60
per M tokens
MoonshotAI
Kimi K3
$4.00 / $18.00
per M tokens
DeepSeek
DeepSeek V4.1 Flash
$0.40 / $1.40
per M tokens
/jab-hook — Jab-jab-jab-right-hook for a personal portfolio
Gary Vaynerchuk's framework applied to your configured portfolio: keep the feed mostly jabs (value — BIP + educational) so the hooks (promo) earn attention when they land. Drafts to your personal Typefully workspace.
One-time setup: see references/properties.md and references/typefully-config.md to configure your real portfolio + workspace. Personal config lives in ${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/jab-hook/.
Mental model
- N rotation slots (configured in
~/.config/makerskills/jab-hook/properties.yaml): one per property you want to promote on rotation - ~2 promo posts/week → each property cycles every ~3 weeks at equal weight
- Non-promo days: 50/50 build-in-public vs educational
- Target weekly rhythm: ~2 promo + ~2–3 BIP + ~2–3 educational
- Platforms: X + LinkedIn via your personal Typefully workspace
Step 1 — Pick the mode
Detect from the trigger:
| User says | Mode |
|---|---|
| "plan my socials," "next week's posts" | plan |
| "what should I post," "next promo," "pick next" | pick-next |
| "audit my socials," "what's overdue," "what haven't I posted about" | audit |
| "draft a promo for X," "BIP post about Y," "educational about Z" | draft |
If ambiguous, confirm.
Step 2 — Load context
- Read
references/properties.md— the 6 rotation slots and angle ideas - Read
references/voice.md— the user's voice rules per platform - Read
references/content-types.md— templates for promo / BIP / educational - Pull recent posts from your personal Typefully workspace:
- First run: call
mcp__typefully__typefully_list_social_setsand ask which social set is your personal (X + LinkedIn). Save the ID to${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/jab-hook/typefully.yamlfor future runs (schema inreferences/typefully-config.example.yaml— never save it inside the skill folder; upgrades wipe it). - Call
mcp__typefully__typefully_list_draftsfiltered to the last 30 days - Classify each as promo / BIP / educational by content
- First run: call
- Compute days since last promo for each of the 6 slots
Step 3 — Mode logic
plan (7-day plan)
- Slot the 1–2 most overdue properties as promo posts
- Fill remaining 5 days with alternating BIP and educational (50/50)
- Output as a table:
Day | Type | Property/Topic | Hook | Draft summary - Don't stack two promo on consecutive days
- After the user approves the plan, offer to draft each one in sequence
pick-next
- Identify the most overdue property
- If most overdue < 14 days, suggest a BIP or educational instead (whichever is less recent)
- Draft 1 post; ask before syndicating
audit
- Report: days since last promo per property, ordered by most overdue
- Flag any property >21 days as overdue
- Suggest the next 1–2 moves
draft
- Skip rotation logic — draft the requested post
- Use the right content-type template
Step 4 — Draft
Follow references/voice.md and references/content-types.md. Default: single post, one per platform (X version + LinkedIn version). Thread only if the second post earns its place.
Link placement (load-bearing): never put a URL in the body. Default is no link at all — let curiosity drive the click. When a link is needed, draft it as a first comment for LinkedIn (Typefully supports auto-comment) or a reply for X. The body always stands alone.
Optional: inspiration scan
For promo and educational posts, optionally scan inspiration accounts (references/inspiration.md) before drafting:
- Pick 1–2 accounts whose audience overlaps with the property
- Pull recent posts (agent-browser if it works on LinkedIn; otherwise ask the user to paste examples or fall back to their X/newsletter)
- Extract structural patterns: hook openers, post length, line-break rhythm, CTA styles
- Apply ONE pattern to the user's voice — never mimic phrasing
- Log new patterns in
${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/jab-hook/patterns.local.md(create if missing) so they accumulate —references/patterns.mddocuments the starter patterns and stays read-only
Skip inspiration scan when the user says "just draft it" or when iterating on an existing post.
Step 5 — Syndicate to Typefully
Ask: "Push to Typefully now? (X + LinkedIn, your personal workspace)"
If yes:
- Read social set ID from
${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/jab-hook/typefully.yaml(setup notes inreferences/typefully-config.md) - Call
mcp__typefully__typefully_create_draftonce for X, once for LinkedIn (or once with both platforms if the social set spans both) - If the draft has a first-comment link, configure Typefully to add it automatically (LinkedIn auto-comment / X reply). If the MCP doesn't support auto-comment for one of the platforms, surface the comment text so the user can paste it after publishing.
- Default to draft state (not scheduled) — the user reviews in Typefully UI before sending
- Return the Typefully draft URLs
Composes with
paste— clean output for pasting into Typefully / LinkedIn / X manually if the MCP path doesn't fitsocial-fetch— pull inspiration-account posts for structural analysis (replaces inline agent-browser)second-brain— theContent Ideaswiki page hoppers hooks, frameworks, stories;/jab-hookdrafts pull candidates from theredeep-research— when a promo needs a stat or citation the draft doesn't have yetmarketingskills:social— generic social frameworks, useful when teaching strategy vs shipping your own postsmarketingskills:copywriting— for hook / headline ideation when stuck
Notes on quality
- 1 promo per week per property is the floor, not the ceiling. Rotation compounds when it's predictable — 4 properties × 1 promo/week = 4 promo weeks/month. Educational + BIP + community fill the rest.
- Never mimic phrasing from inspiration accounts. Extract structure (hook opener type, line-break rhythm, CTA style) and apply to your own voice. Voice is the moat; copying phrasing destroys it.
- Links belong in first comments, not the body for LinkedIn + X. Documented in
~/.claude/memory/feedback_social_link_placement.md. Body posts with inline URLs get algorithm-suppressed on both platforms. - Draft state, not scheduled state. Every Typefully push defaults to draft — human review in the Typefully UI before publish is non-negotiable. Automated scheduling of unreviewed drafts has repeatedly produced posts the user regrets.
- Reader-perspective framing. "You'll know X" not "we'll teach you X." "What you're guaranteed" not "what we'd guarantee."
- Voice.local.md overrides the shipped voice.md. Personal voice rules never live in the public repo — always in
${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/jab-hook/voice.local.md. - Cadence cap: 2 posts/day per platform, per account. More than that degrades reach + reads as spam.
Memory references
~/.claude/memory/feedback_promo_voice.md(if present) — conviction-coded CTAs + reader-perspective framing~/.claude/memory/feedback_social_cadence.md(if present) — cadence cap (e.g., 2 posts/day per platform)
Installationen
Sicherheitsprüfung
Quellcode ansehen
coreyhaines31/makerskills
Mehr aus dieser Quelle
Give agents clean web context
Search and extract the public web as Markdown or structured JSON through one API or hosted MCP server.
Verwandte Skills
Mehr aus dieser Quelle: coreyhaines31/makerskills
So verwenden Sie diesen Skill
Install jab-hook by running npx skills add coreyhaines31/makerskills --skill jab-hook in your project directory. Führen Sie den obigen Installationsbefehl in Ihrem Projektverzeichnis aus. Die Skill-Datei wird von GitHub heruntergeladen und in Ihrem Projekt platziert.
Keine Konfiguration erforderlich. Ihr KI-Agent (Claude Code, Cursor, Windsurf usw.) erkennt installierte Skills automatisch und nutzt sie als Kontext bei der Code-Generierung.
Der Skill verbessert das Verständnis Ihres Agenten für jab-hook, und hilft ihm, etablierte Muster zu befolgen, häufige Fehler zu vermeiden und produktionsreifen Code zu erzeugen.
Was Sie erhalten
Skills sind Klartext-Anweisungsdateien — kein ausführbarer Code. Sie kodieren Expertenwissen über Frameworks, Sprachen oder Tools, das Ihr KI-Agent liest, um seine Ausgabe zu verbessern. Das bedeutet null Laufzeit-Overhead, keine Abhängigkeitskonflikte und volle Transparenz: Sie können jede Anweisung vor der Installation lesen und prüfen.
Kompatibilität
Dieser Skill funktioniert mit jedem KI-Coding-Agenten, der das skills.sh-Format unterstützt, einschließlich Claude Code (Anthropic), Cursor, Windsurf, Cline, Aider und anderen Tools, die projektbezogene Kontextdateien lesen. Skills sind auf Transportebene framework-agnostisch — der Inhalt bestimmt, für welche Sprache oder welches Framework er gilt.