Channel Post Generator

X (Twitter)

• Length: 280 chars per tweet, 8 max in thread
• Hook: First line IS the hook — no preamble
• Tone: Sharp, direct, high-signal
• Structure: Single tweet or 3-5 tweet thread
• Hashtags: 1-3 max, integrated naturally
• NO: Hashtag dumping, engagement farming, "1/" thread numbering unless long

Instagram

• Length: 150-300 chars caption (2200 max)
• Hook: First line grabs attention in feed preview
• Tone: Visual-first, conversational but professional
• Structure: Short caption + CTA + hashtags block
• Hashtags: 5-15 relevant, in separate block
• Requires: Image brief must be generated alongside

Discord

• Length: 2000 chars max
• Tone: Community-friendly, direct, helpful
• Structure: Short update with key info + link
• NO: Marketing-speak, @everyone (unless launch)

Reddit

• Length: Varies by subreddit
• Tone: Useful, discussion-oriented, zero hype
• Structure: Value-first, question-ending
• NO: Self-promotion language, brand mentions without context
• REQUIRES: L3 approval always

Two-Stage Output

Stage 1: Internal Planning (not published)

{
  "platform": "linkedin",
  "product": "testsprep",
  "content_type": "educational",
  "goal": "authority",
  "hook_options": ["option A", "option B"],
  "cta_options": ["Visit testsprep.in", "Try practice questions"],
  "source_links": ["https://testsprep.in"],
  "claims_used": ["AI-powered test prep for Indian exams"],
  "restricted_claims_avoided": ["guaranteed results"],
  "image_brief": "Clean dashboard showing study progress..."
}

Stage 2: Final Output (this is what gets posted)

Clean, human-readable text ONLY. No JSON. No metadata. No field names.

Writing Rules

1. Lead with value — what does the reader get?
2. Use active voice
3. Short sentences, short paragraphs, scannable
4. No filler: "In today's fast-paced world...", "It goes without saying..."
5. Specific over vague: numbers, examples, concrete details
6. Honest about limitations — never oversell
7. Builder perspective — we make things, we ship things, we learn things
8. No banned phrases from brand-voice-rules.json

Quality Checks

• Post text passes sanitize_post.py validation (no JSON/metadata leaks)
• Every claim is in product truth file's "Safe Claims" section
• No restricted claims used without explicit flagging
• Length is correct for platform
• Tone matches platform rules
• CTA is relevant and not aggressive
• Hashtags are within platform limits
• No banned phrases from brand-voice-rules.json

HONEST CLASSIFICATION: This is a prompt template, not an executable plugin. OpenClaw injects this as context to guide LLM behavior. It does NOT enforce rules at runtime.