agentleFS
fortunto2fortunto2/solo-factory/skills/content-gen/SKILL.md

solo-content-gen

Skill18 starsChanged 4 months agoMITView on GitHub

Use when "create content", "write LinkedIn post", "social media pack", "marketing copy", "promotion content", "posts from this article", "repurpose this post", "посты из статьи", or need social posts from a PRD or from a published write-up. Do NOT use for community thread responses (/community-outreach) or video scripts (/video-promo).

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 12 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When the input is a post that is already published (the canonical write-up on the user's own site), the job is distribution, not …

    line 20
  2. Distinctive:

    Facts only from the write-up: Every number in a draft appears in the source post.

    line 30
  3. Distinctive:

    If MCP tools are not available, fall back to Glob + Grep + Read.

    line 48
  4. Very specific:

    If project name: find PRD at docs/prd.md, or README.md, or CLAUDE.md.

    line 53
  5. Distinctive:

    If path: read directly.

    line 54
  6. Distinctive:

    If empty: ask via AskUserQuestion — "Which project to generate content for?"

    line 55
  7. Distinctive:

    Before generating content, write out:

    line 66
  8. Distinctive:

    Use line breaks for readability

    line 78
  9. Distinctive:

    NOT promotional — value-first, product mention is secondary

    line 83
  10. Distinctive:

    Write as if answering someone's genuine question

    line 85
  11. Distinctive:

    If commits exist, generate user-facing release notes:

    line 101
  12. Distinctive:

    Keep tone consistent with ICP (casual for consumer, professional for B2B)

    line 105
  13. Distinctive:

    All content is draft — always review before publishing

    line 162
  14. Very specific:

    Fix: Ensure PRD has detailed problem statement, ICP persona, and unique angle.

    line 184

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/content-gen

Generate a promotion content pack from a project's PRD or README. Produces 3 ready-to-use content pieces: LinkedIn post, Reddit answer draft, and Twitter/X thread. For video scripts, use /video-promo.

Mode: posts from a published write-up

When the input is a post that is already published (the canonical write-up on the user's own site), the job is distribution, not copywriting from scratch. Each channel gets its own rewrite, never a copy:

  • One contract per channel, kept as data the user can edit (voice, audience, language, length limit, where the link goes, which media it needs, cadence, how it gets out: scheduled queue, a draft the user posts, or an upload after an OK). Read it before writing; do not carry a voice in your head from the last post.
  • Write for that channel's audience. A professional network wants the lesson and the number; a build-in-public feed wants the story and the surprise; a channel of strangers wants the general insight with the user's work as one example inside it, not a diary entry.
  • Facts only from the write-up. Every number in a draft appears in the source post. A claim the post does not make does not appear in the derivative.
  • One link, tagged. The canonical post URL with utm_source=<channel>&utm_medium=social&utm_campaign=<slug>, placed where the channel wants it (end of a Telegram post, LinkedIn's first comment rather than the body, the last part of an X thread).
  • If the project has a gated contract tool (a brief that assembles the full context and a draft that validates against the contract), use it and nothing else: brief, write, draft, fix until every gate passes. A post that does not fit a channel is skipped with a written reason, never left silently undone.
  • Never post from the user's accounts unless the contract says that channel's mechanics allow it; drafts for the user are the default.

MCP Tools (use if available)

  • kb_search(query) — find related methodology and principles
  • project_info(name) — get project details, stack, description
  • symbol_search("<words>") — find key features in codebase (symbol names, every project; Grep inside one)

If MCP tools are not available, fall back to Glob + Grep + Read.

Steps

  1. Parse project from $ARGUMENTS.
    • If project name: find PRD at docs/prd.md, or README.md, or CLAUDE.md.
    • If path: read directly.
    • If empty: ask via AskUserQuestion — "Which project to generate content for?"
  2. Extract core messaging from PRD/README (parallel reads):
    • Problem: What pain does this solve? (1 sentence)
    • Solution: What does the product do? (1 sentence)
    • ICP: Who is the target user? (persona)
    • Key features: Top 3-5 differentiating features
    • Unique angle: What nobody else does (gap from research.md if exists)
    • Social proof: Any metrics, users, downloads (if available)
  3. Forced reasoning — messaging framework: Before generating content, write out:
    • Hook: What stops the scroll? (pain point or surprising stat)
    • Proof: Why should they believe you? (demo, numbers, authority)
    • CTA: What's the one action? (try, sign up, download)
    • Tone: Based on ICP — casual/professional/technical/friendly
  4. Generate content pieces:

    4a. LinkedIn Post

    Format: Hook line → Story (before/after) → Insight → CTA

    • 1300-1500 chars max
    • First line is the hook (visible before "see more")
    • Use line breaks for readability
    • End with question to drive engagement

    4b. Reddit Answer Draft

    Format: Helpful answer that naturally mentions the product

    • NOT promotional — value-first, product mention is secondary
    • Suggest 3-5 subreddits where this would be relevant
    • Write as if answering someone's genuine question
    • Include "disclaimer: I built this" transparency

    4c. Twitter/X Thread (5-7 tweets)

    Format: Hook tweet → Problem → Solution → Demo → Social proof → CTA

    • Each tweet: 240-280 chars max
    • Tweet 1: standalone hook (works without thread)
    • Last tweet: CTA with link
    • Include emoji sparingly (1-2 per tweet max)
  5. Generate Release Notes (if project has git history):

    git log --oneline --since="1 month ago" 2>/dev/null | head -20

    If commits exist, generate user-facing release notes:

    • Group by: New features, Improvements, Bug fixes
    • Translate technical commit messages into user language
    • Skip internal commits (chore, refactor, test, docs)
    • Keep tone consistent with ICP (casual for consumer, professional for B2B)

    Format:

    ## What's New (v{version})
    
    **New**
    - {user-facing feature description}
    
    **Improved**
    - {improvement in user terms}
    
    **Fixed**
    - {bug fix in user terms}

    Include in content pack. If no git history, skip this section.

  6. Write content pack to docs/content-pack.md:

    # Content Pack: {Project Name}
    
    **Generated:** {YYYY-MM-DD}
    **ICP:** {target persona}
    **Core message:** {one-line value prop}
    
    ## Messaging Framework
    - **Hook:** {hook}
    - **Proof:** {proof}
    - **CTA:** {cta}
    
    ---
    
    ## 1. LinkedIn Post
    {ready to copy-paste}
    
    ## 2. Reddit Draft
    **Target subreddits:** r/..., r/..., r/...
    {answer draft}
    
    ## 3. Twitter/X Thread
    {numbered tweets}
    
    ## 4. Release Notes
    {release notes from step 5, or "No git history — skipped"}
    
    > For video script, run `/video-promo {project}`.
    
    ---
    *Generated by /content-gen. Edit before publishing.*
  7. Output summary — print core message + list of generated content pieces.

Notes

  • All content is draft — always review before publishing
  • Reddit draft should feel genuinely helpful, not promotional
  • For video scripts, use /video-promo (owns video generation entirely)
  • Rerun after major product updates to refresh messaging
  • Works best with detailed PRD — falls back to README if no PRD

Gotchas

  1. LinkedIn algorithm penalizes external links — links in the post body reduce reach by 40-50%. Put the link in the FIRST COMMENT instead, and mention "link in comments" in the post.
  2. Twitter/X thread hooks must work standalone — tweet 1 appears alone in feeds without the thread. If it doesn't grab attention by itself, no one clicks "Show this thread".
  3. Reddit self-promotion rules — most subreddits enforce a 10:1 ratio (10 helpful comments per 1 promotional post). A brand-new account posting product links gets instantly removed.
  4. AI writing patterns are detectable — em dashes, "game-changer", "dive into", rule-of-three lists, "In today's fast-paced world". Run output through /humanize before publishing.
  5. Release notes ≠ changelog — user-facing release notes should describe BENEFITS ("faster checkout"), not technical changes ("refactored payment module"). Skip internal commits entirely.

Common Issues

No PRD found

Cause: Project doesn't have docs/prd.md or README. Fix: Provide a project path with PRD, or run /validate first to generate one.

Content too generic

Cause: PRD lacks specific differentiators or ICP details. Fix: Ensure PRD has detailed problem statement, ICP persona, and unique angle. Add research.md for competitive context.

Wrong tone for audience

Cause: ICP not clearly defined in PRD. Fix: Specify the target audience in PRD. LinkedIn = professional, Reddit = casual/technical, Twitter = punchy/conversational.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.