4 · LinkedIn Profile

HIFI PROTOTYPE · Section 4 · Tier 1live

Purpose: Render the founder's LinkedIn profile — headline, about, how-it-works, CTA — in branded voice.

Inputs → Output

Sources

SPP2

Reads

block_eblock_c.social_proofblock_a.brand_namesection_24

Writes

linkedin_profile

Pipeline

Load anchors

SP loaded as system message (voice anchor, Rule 4). Task prompts: SP · P2. Pull ≥3 reference IDs (Rule 2). Reads locked upstream: 24.

Generate

Headline ≤220ch · About ≤2000ch · CTA

Enforce caps

Headline ≤220ch · About ≤2000ch.

QA gate

Section 25 auto-runs QA L1 (24 checkpoints) + L2 (6 brand gates). Any L2 fail = REJECT → rewrite.

Render + ship

Sync to brand.assets. POC approves in portal → /api/brand-os/approve writes the asset → ships to domain.

Technical Reference

Generation contract (machine-executable)

section: "4"
id: s04-linkedin-profile
state: HIFI_PROTOTYPE
tier: 1
sources: ["SP", "P2"]
inputs:
    - block_e
    - block_c.social_proof
    - block_a.brand_name
    - section_24
output: "Headline ≤220ch · About ≤2000ch · CTA"
caps:
    - Headline ≤220ch
    - About ≤2000ch
cascades_from: [24]
cascades_to: []

Supabase sync map (from Part 10)

Section 4 ->
  brand.assets where type='linkedin_profile'

SCHEMA NOTE: verify all jsonb paths + brand.assets type values against live Supabase before write. Flag missing columns in the PR; never create new columns without review.

Note

This is a HIFI artifact: its consumer is a real customer. It only renders content that already passed the LOFI gate upstream — it never invents facts. Source of truth: Master Template Part 3 + the Master Index. To change this page, edit content/docs/roadmap/_sections.json and re-run the compiler.