gi-brief-extractor

name	gi-brief-extractor
description	Extract structured briefs from GreatInvert source material (Playbook chapters, blog drafts, transcripts, notes). Use when Tim says "brief this," "extract a brief," "structure this source," or drops source content and wants it turned into a reusable messaging brief for the asset pipeline. Produces a completed Structured Brief with audience, problem, promise, takeaways, objections, CTAs, and visual metaphors, then writes it to the tracker spreadsheet and saves a Markdown document.
metadata	{"author":"tim-dickey","version":"1.0","project":"greatinvert"}

GreatInvert Structured Brief Extractor

When to Use This Skill

Use when the user:

• Says "brief this," "extract a brief," or "structure this"
• Drops source material (Playbook chapter, blog draft, transcript, notes) and wants it processed through the asset pipeline
• Asks to populate the Structured Briefs tab of the tracker
• Wants to turn raw content into reusable messaging blocks

Prerequisites

• The GreatInvert asset pipeline tracker spreadsheet (.xlsx with a "Structured Briefs" tab)
• Source material to extract from (text, document, or file content)

Instructions

Step 1: Locate the tracker

Find the tracker spreadsheet. Check these locations in order:

1. User's workspace: look for *pipeline*tracker*.xlsx or *greatinvert*tracker*.xlsx
2. Google Drive: search for "greatinvert-asset-pipeline-tracker" using the files connector
3. Ask the user where their tracker is

Store the local path for use in Step 5.

Step 2: Read the source material

Read the full source content. This may be:

• A file the user attached or pointed to
• Content from a Google Drive document
• Text pasted directly in the conversation
• A URL to fetch

Get the complete text — do not truncate. If the source is long (e.g., a full Playbook chapter), read it all. The quality of the brief depends on having the full context.

Step 3: Extract the structured brief

Read the schema reference at references/brief-schema.md for field-level guidance.

Analyze the source material and extract every field. Apply these rules:

Audience: Be specific about the professional role and context. "Product managers navigating AI adoption in enterprise teams" not "product people."

Funnel stage: Match to where this content sits in the buyer journey. Playbook chapters are typically Consideration. Launch announcements are Conversion. Thought leadership is Awareness.

Core Problem / Promise / Key Insight: These three form the editorial backbone. The problem should sting. The promise should feel attainable. The insight should be surprising or counterintuitive.

Takeaways (3): These become social posts, quote cards, and section headers. Each must stand alone. Keep under 12 words.

Objections (2): Write in the skeptic's voice. These drive FAQ content and objection-handling copy.

CTAs: Primary matches funnel stage. Secondary is a lower-commitment step.

Visual Metaphors (2): Conceptual, not literal. Must align with Great Inversion brand — editorial, strategic, grounded. Never suggest robots, circuit boards, glowing orbs, or generic AI imagery.

Tone Notes: Reference the brand voice — editorial, grounded, practical, strategic — and note any specific tonal nuances from this particular source.

Priority: High if it directly supports the Playbook or an active campaign. Medium for upcoming releases. Low for future/archival.

Approved: Always "No" — the human approves after review.

Output the brief as a JSON object with all 19 fields.

Step 4: Present the brief for review

Display the extracted brief to the user in a clean readable format before writing anything. Let them see all fields and confirm or request changes. Show it as a table or structured list, not raw JSON.

Step 5: Write to the tracker spreadsheet

After the user confirms (or immediately if they've indicated they don't need to review):

1. Save the brief JSON to a temp file
2. Run the tracker update script:

python skills/gi-brief-extractor/scripts/update_tracker.py \
  <tracker_path> \
  /tmp/brief_data.json \
  --source-json /tmp/source_data.json  # if source isn't already in the tracker

3. The script auto-generates the Brief_ID and appends the row.

Step 6: Save the brief document

Save a readable Markdown version of the brief:

python skills/gi-brief-extractor/scripts/save_brief_doc.py \
  /tmp/brief_data.json \
  /home/user/workspace/pipeline-docs/02-Structured-Briefs/<brief-type>/BRF-XXX_<slug>.md

Choose the subfolder based on content:

• Product-Briefs/ — briefs about the Playbook or other products
• Chapter-Briefs/ — briefs from individual Playbook chapters
• Campaign-Briefs/ — briefs for specific campaigns or launches
• Visual-Direction/ — briefs focused on visual identity

Step 7: Upload to Google Drive

If the Google Drive connector is available:

1. Export the updated tracker spreadsheet back to Drive (replacing the existing file)
2. Export the brief Markdown to the matching 02-Structured-Briefs/ subfolder in Drive

Step 8: Confirm completion

Tell the user:

• The Brief_ID assigned
• Where the brief document was saved
• That the tracker was updated
• That the brief is marked "Approved: No" and needs their review to flip to Yes
• Suggest next steps: "You can now use this brief to generate assets in Step 3 of the pipeline"

Multiple Briefs from One Source

A single source document (especially a Playbook chapter) can yield multiple briefs — one per audience segment, funnel stage, or channel angle. When the source is rich enough:

1. Ask the user if they want a single brief or multiple angles
2. If multiple, extract each as a separate brief with its own Brief_ID
3. Each brief should have a distinct Audience, Funnel_Stage, and editorial angle
4. Run the tracker script once per brief

Quality Checklist

Before presenting the brief, verify:

• Every field has a substantive value (no blanks, no "N/A")
• Core Problem / Promise / Insight form a coherent editorial angle
• Takeaways are concise enough for social posts (under 12 words each)
• Visual metaphors avoid generic AI imagery
• CTAs match the funnel stage
• Tone notes reference the Great Inversion brand voice
• The brief is self-contained — someone could write copy from it without reading the source