post-writing
Skill1 starsChanged 6 months agoApache-2.0View on GitHub
Use when the user asks to write, draft, or rewrite a post for social media (LinkedIn, Telegram, Blog) or wants content to be more engaging.
In this repository23 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- brainstormingSkill
- design-mdSkill
- docxSkill
- hooks-creatorSkill
- htmlSkill
- marp-slideSkill
- mcp-builderSkill
- obsidian-cliSkill
- pdfSkill
- pptxSkill
- skill-auto-improveSkill
- skill-creatorSkill
- skill-enhancerSkill
- skill-validatorSkill
- summarizing-meetingsSkill
- text-humanizerSkill
- transcript-fetcherSkill
- vdd-adversarialSkill
- vdd-sarcasticSkill
- wiki-ingestSkill
- xlsxSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
18 rules. About 0 are common, 11 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
Distinctive · estimate
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 52
Write scope: Create or modify ONLY the post file the user asked for.
- Distinctive: line 55
No borrowed experience: Never fabricate first-person stories, failures, launches, or results the user has not claimed.
- Very specific: line 60
Primary: Every item in references/review-checklist.md — the Core block and the Advanced block — passes before the draft reaches the user.
- Distinctive: line 77
If ANY of these are missing, ASK the user first.
- Very specific: line 81
Read references/hooks-examples.md for hook types and formulas.
- Distinctive: line 83
ASK the user to choose one.
- Distinctive: line 89
Read the appropriate voice guide:
- Very specific: line 90
RU posts: Read references/voice-guide-ru.md
- Very specific: line 91
EN posts: Read references/voice-guide-en.md
- Very specific: line 92
Other languages: Use references/voice-guide-en.md as fallback.
- Very specific: line 93
Read references/platform-rules.md for platform-specific formatting.
- Distinctive: line 105
Use Data + Narrative + Framework layers for maximum reach.
- Distinctive: line 108
DO NOT start with "In this post I will talk about..."
- Distinctive: line 109
DO NOT use filler. Every sentence MUST earn its place.
- Distinctive: line 110
Use "You" and "I" — make it personal.
- Distinctive: line 115
Verify the draft against every checklist item.
- Distinctive: line 116
If ANY check fails — fix immediately before presenting to the user.
- Distinctive: line 127
Default paths (if the user specifies a different location, use that instead):
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Post Writing Skill
When to Use
- User asks to write a post for LinkedIn, Telegram, or Blog.
- User wants to rewrite text to be more engaging or "viral".
- User mentions copywriting techniques (hooks, slippery slide, etc.).
Red Flags (Anti-Rationalization)
STOP and READ if you are thinking:
- "I'll just write something generic" → WRONG. Every post MUST have a specific hook, personal story, and concrete data.
- "The user didn't specify the audience, I'll assume" → WRONG. ASK first. Audience determines the hook type and awareness level.
- "This hook is good enough" → WRONG. Generate 3 options. Let the user choose.
- "I'll skip the review checklist" → WRONG. EVERY draft MUST pass the checklist before presenting.
- "I'll load all resources at once" → WRONG. Load ONLY the resources needed for the current step.
Rationalization Table
| Agent Excuse | Reality |
| "The topic is straightforward, no hook needed" | Even simple topics need hooks — flat openings kill engagement |
| "I already know the style, no need to read voice guide" | Voice guides contain specific lexicon and anti-patterns unique to this brand |
| "Templates are optional for experienced writers" | Templates ensure structural consistency across posts |
| "The checklist is redundant, I reviewed mentally" | Mental reviews miss items. Use the explicit checklist every time |
Execution Mode
- Mode:
prompt-first - Rationale: This skill ships no
scripts/directory and no runtime dependency. Its output is prose. Hook choice, voice match, and platform formatting are judgement calls made by readingreferences/and writing Markdown — there is nothing here for an executable to decide.
Script Contract
This skill ships no executable. The contract it honours is the file it writes; the values below restate ## File Conventions as a checkable interface.
- Input: the user's brief plus the Step 1 answers (goal, audience, key takeaway, platform, language).
- Output: exactly one Markdown file per requested post.
- Naming:
YYYY-MM-DD Post Title.md— ISO date, space, human title. - Destination:
Draft Posts/RU/orDraft Posts/EN/under the project path listed in## File Conventions. A location named by the user overrides that default. - Format: Markdown body matching the target platform — → arrows and no headings for LinkedIn, full Markdown for Telegram.
- Promotion: a draft moves to
Published Posts/ONLY on an explicit user instruction.
Safety Boundaries
- Write scope: Create or modify ONLY the post file the user asked for. The skill's own
assets/andreferences/are read-only inputs — never edit, extend, or overwrite them during a drafting run. - No publishing: This skill writes a file and stops. It never posts, schedules, or transmits content to LinkedIn, Telegram, a blog engine, or any API. Publication stays a human action.
- No invented facts: Statistics, quotes, customer names, company names, dates, and metrics come from the user. When a hook formula needs a number the user has not supplied, ASK for it. Never fill the slot with a plausible-looking figure.
- No borrowed experience: Never fabricate first-person stories, failures, launches, or results the user has not claimed. "Embed authority through details" means the user's details.
- Confidentiality: Personal and client information the user supplies stays inside the requested artifact — never copied into the skill's own references, never repeated into unrelated files.
Validation Evidence
- Primary: Every item in
references/review-checklist.md— the Core block and the Advanced block — passes before the draft reaches the user. - Secondary: Length and formatting match the target platform's limits in
references/platform-rules.md(LinkedIn 800-1,500 characters, → arrows, no headings; Telegram 500-2,000 characters, full Markdown). - Hook gate: The selected hook clears the Step 2 hook checklist in
references/hooks-examples.md— specific, honest, open loop, matched to the reader's awareness level. - Inspectable: A caller confirms the pass by opening the saved
YYYY-MM-DD Post Title.mdin the drafts folder and reading it against those two checklists; worked end-to-end traces live inexamples/.
Workflow
Step 1: Analyze & Clarify
Identify before writing:
- Goal — sales, engagement, education, or brand awareness?
- Target Audience — Unaware, Problem-Aware, Solution-Aware, Product-Aware, or Most Aware? (use Schwartz's 5 awareness levels)
- Key Takeaway — What MUST the reader do/think after reading?
- Platform — LinkedIn or Telegram? (determines formatting rules)
- Language — RU or EN? (determines voice guide)
If ANY of these are missing, ASK the user first. DO NOT proceed without clarity.
Step 2: Select a Hook
- Read
references/hooks-examples.mdfor hook types and formulas. - Propose 3 distinct hook options (e.g., one Story, one Problem, one Contrarian Thesis).
- ASK the user to choose one.
Apply Ogilvy's principle: "The headline is 80 cents of your dollar." Spend time on the hook.
Step 3: Draft the Post
- Read the appropriate voice guide:
- RU posts: Read
references/voice-guide-ru.md - EN posts: Read
references/voice-guide-en.md - Other languages: Use
references/voice-guide-en.mdas fallback. Note to the user that no brand-specific voice guide exists for this language.
- RU posts: Read
- Read
references/platform-rules.mdfor platform-specific formatting. - Optionally read
assets/templates.mdif the user wants a specific format (listicle, structured, multi-platform).
Structure (Problem → Reframe → Action):
- Hook — contradiction, intrigue, or personal story
- Problem — what's wrong with the current situation (personal experience, specifics)
- Reframe — why common wisdom is incomplete (your perspective from experience)
- Action — what you're doing about it + invitation to dialogue
Drafting Rules:
- Apply Sugarman's Slippery Slide: every sentence's sole purpose is to make the reader read the next one.
- Match the hook to audience awareness level (Schwartz).
- Use Data + Narrative + Framework layers for maximum reach.
- Specific numbers in the first lines (not "many" but "776 professionals").
- Embed authority through details, DO NOT claim it.
- DO NOT start with "In this post I will talk about..."
- DO NOT use filler. Every sentence MUST earn its place.
- Use "You" and "I" — make it personal.
Step 4: Review
- Read
references/review-checklist.md. - Verify the draft against every checklist item.
- If ANY check fails — fix immediately before presenting to the user.
File Conventions
Naming
- Format:
YYYY-MM-DD Post Title.md - Examples:
2026-02-11 Нашёл ассистента.md,2026-02-05 Vibe-coded mobile Claude Code in one evening.md
Paths
Default paths (if the user specifies a different location, use that instead):
- Drafts RU:
01_Projects/HowToAI - blog buildinpublic/Draft Posts/RU/ - Drafts EN:
01_Projects/HowToAI - blog buildinpublic/Draft Posts/EN/ - Published RU:
01_Projects/HowToAI - blog buildinpublic/Published Posts/RU/ - Published EN:
01_Projects/HowToAI - blog buildinpublic/Published Posts/EN/
Saving Rules
- Save drafts to
Draft Posts/RU/(orEN/) by default. - Move to
Published Posts/ONLY when the user explicitly says to publish.
Resources
- Hook Types & Examples — formulas and examples for 7 hook types
- Voice Guide — RU — Russian voice, tone, lexicon, anti-patterns
- Voice Guide — EN — English voice, tone, anti-patterns
- Platform Rules — LinkedIn vs Telegram formatting
- Post Templates — post format templates
- Review Checklist — pre-publish verification
Examples
- Worked Example — EN, LinkedIn — full four-step trace ending in a Contrarian Thesis hook, with four checklist failures and the three revisions they forced
- Worked Example — RU, Telegram — full four-step trace ending in a Story Lead hook, with the "Раньше/Сейчас" framework added at review
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

