b2bforce LinkedIn Post Generation
Draft pipeline for a content idea of type linkedin_post.
When to Use
- Idea with
content_type: linkedin_post - User wants LinkedIn-native copy (not cross-posted blog summary)
Prerequisites
- Content idea file + firm/service/ICP/persona context
- Current agent/LLM session
- Do NOT run Exa reference search (skipped for this type)
Prompt layering
- Base writer (style only)
- Quality layer
- LinkedIn format layer →
references/format-requirements.md
Optional: workspace/firm/content-guidelines/linkedin-post.md
Follow the shared draft gate and writing quality rules in docs/content-generation.md.
Use buyer_question, unique_angle, proof_source, next_action, and
research_mode from the idea; do not write from title + description only.
Output format (critical)
- Plain text — NOT Markdown headers
- One sentence per line
- 150–250 words optimal; max ~300 words / 3000 chars
- 0–2 emojis if brand-appropriate
- 2–3 hashtags at end
- Engagement question at end
- No bullet lists (hurts engagement)
- First-person voice (I, we, you)
Stage → framework (from format prompt)
| Stage | Framework |
|---|---|
| Problem | Story Hook, Poll |
| Concept | Category, This-vs-That |
| Education | Framework, How-To |
| Decision | Comparison, Data |
| Vendor | CSR, Behind-the-Scenes |
Output path
Brand scope. In a multi-brand workspace the draft inherits the idea file's brand: write to workspace/marketing/content/{brand}/drafts/… — same brand segment as the idea. See the Brand Scope Gate in AGENTS.md.
workspace/marketing/content/drafts/linkedin/{slug}.md
Frontmatter: idea, content_type, buying_stage, service, icp, persona,
generated_at. Body = plain text only. Title in frontmatter for filing.
Run scripts/validate-content-draft.sh {output}. Update idea frontmatter to
status: generated after the draft validates.
Related Skills
marketing-content-x-post — different length/platform rules