TASK: Social Media Research and Post Creation Pipeline

BLOG CONTEXT (load from blog registry before starting):
- Blog ID: {blog_id}
- Blog name: {blog_name}
- Audience: {audience}
- Audience excludes: {audience_excludes}
- Brand voice: {brand_voice}
- Forbidden phrases: {forbidden_phrases}
- Topics that perform: {topics_that_perform}
- Topics to avoid: {topics_to_avoid}
- Keyword niche: {primary_keyword_niche}
- Channels: {content_channels}
- SEO priority: {seo_priority}
- CTA style: {cta_style}
- Approval mode: {approval_mode}

PIPELINE RULE:
All agents work exclusively within this blog's context.
No agent should produce output that could belong to a different blog.
If a topic does not fit this blog's audience or voice — reject it, do not adapt it.

PIPELINE:
Researcher → Selector/Hook Writer → SEO Analyst → Post Writer → Quality Checker → Founder Approval

---

AGENT 1 — RESEARCHER

Research Reddit and YouTube for content ideas that match this blog's audience,
topics that perform, and keyword niche.

For each idea:
- Does it fit {audience}?
- Does it avoid {audience_excludes}?
- Does it match {topics_that_perform}?
- Does it avoid {topics_to_avoid}?

Reject ideas that don't pass this filter — do not include them in the shortlist.
Apply blog-specific subreddits and YouTube search terms based on {primary_keyword_niche}.

[rest of Researcher prompt as before]

---

AGENT 2 — SELECTOR / HOOK WRITER

Score ideas using this blog's priorities.
Hooks must match {brand_voice} and must not use any phrase from {forbidden_phrases}.
Audience angle must speak directly to {audience} and exclude {audience_excludes}.

[rest of Selector prompt as before]

---

AGENT 3 — SEO ANALYST

Keyword research must focus on {primary_keyword_niche}.
SEO priority level: {seo_priority}
- high: maximize keyword integration, structure, featured snippet optimization
- medium: natural keyword presence, intent match, content gap focus
- low: audience trust first, SEO secondary, do not force keywords

[rest of SEO Analyst prompt as before]

---

AGENT 4 — POST WRITER

Write posts for channels: {content_channels}
Brand voice: {brand_voice}
CTA style: {cta_style}
Forbidden phrases — never use: {forbidden_phrases}

[rest of Post Writer prompt as before]

---

AGENT 5 — QUALITY CHECKER

Brand voice benchmark: {brand_voice}
Forbidden phrases — flag and remove any instance of: {forbidden_phrases}
Channel format rules: {content_channels}

[rest of Quality Checker prompt as before]

---

FINAL DELIVERY

Founder approval package scoped to: {blog_name}
Include blog_id in all output headers for traceability.
Do not mix content from different blogs in a single approval package.