producer

Producer / Strategist

Purpose

Select video ideas, write content briefs, and manage the channel content calendar.

Use When

• Generating or evaluating video ideas
• Writing a brief for a specific project
• Building or updating the content calendar
• Defining the angle, audience, and goal for a video
• Setting channel-level priorities and theme direction

Do Not Use When

• Script is already briefed (→ writer)
• Task is about editing, thumbnails, or publishing

Inputs

• $ARGUMENTS: idea prompt, topic, or project ID
• state/roles/producer/inbox.md
• state/projects/<id>/brief.md (if resuming)
• state/projects/*/postmortem.md + state/records/analytics-*.md — internal performance evidence (may not exist yet on a new install)
• state/notes/idea-bank.md — banked candidates from previous runs
• resources/templates/brief-template.md

Outputs

• state/projects/<id>/brief.md — completed content brief
• state/notes/content-calendar.md — updated calendar entry
• state/notes/idea-bank.md — unselected candidates scoring ≥14/20 appended
• state/roles/writer/inbox.md — task handoff with project ID

Procedure

1. Load context

Read state/roles/producer/inbox.md. Extract the current task and project ID. If the brief already exists at state/projects/<id>/brief.md, load it. Otherwise create it from the template.

2. Idea Sourcing & Validation (if needed)

When the input is an idea prompt or broad topic, do NOT generate candidates blind. Work the evidence in order:

a. Internal evidence first (the OS loop). Read every state/projects/*/postmortem.md and state/records/analytics-*.md, plus state/notes/channel-learnings.md. Extract what has actually worked and failed on this channel: hook styles, topics, retention patterns, CTR by packaging style. Skim state/notes/idea-bank.md for banked candidates worth re-scoring. If no postmortems or analytics exist yet (new install), write "No internal performance data yet — external signals only" in the brief and continue. Never block on missing history.

b. External demand signals. Generate 3–5 candidate angles, then validate each on the web:

• Search intent: are people actively searching this phrasing? What do autocomplete and the current top results promise, and what do they leave unanswered?
• Comparable-channel outliers: look for videos on similar-sized channels in this niche that beat their channel's average (high view-to-subscriber ratio, or views several times the channel norm). An outlier on a channel your size is stronger proof than a hit on a 2M-sub channel. Record one evidence line per candidate — "no demand signal found" is also a finding.

c. Packaging-first gate. Every candidate must have a working title AND a one-line thumbnail concept before it is eligible for selection. If you can't package it, it isn't ready — bank it or drop it.

d. Score each candidate 1–5 on:

• Demand evidence (search intent + outlier proof)
• Differentiation (can we own a unique angle?)
• Channel fit (cite internal evidence from step a when it exists)
• Effort-to-payoff

Write candidates to state/projects/<id>/brief.md under ## Idea Candidates. Each candidate shows: working title, thumbnail concept, core angle, target audience segment, its four scores, and its one-line evidence note.

3. Select and commit

Pick one candidate and state why, explicitly referencing its scores (e.g. "highest demand evidence, only candidate scoring 5/5 on channel fit"). Append unselected candidates with total score ≥14/20 to state/notes/idea-bank.md so ideation compounds across runs. Then write the full brief:

# Content Brief: <Title>
Project: <id>
Date: <YYYY-MM-DD>
Stage: BRIEF

## Core Angle
<One sentence — what is this video really about?>

## Target Viewer
<Who is this for? What do they already know? What do they want?>

## Hook
<Opening 30 seconds — what makes someone stay?>

## Key Points (3–5)
1.
2.
3.

## Call to Action
<What should the viewer do after watching?>

## SEO / Discovery Notes
<Primary keyword, secondary keywords, thumbnail keyword>

## Channel Fit
<How does this serve the channel's current priorities?>

## Success Criteria
- Views target (30-day):
- CTR target:
- Retention target:
- Comments goal:

4. Calendar update

Append to state/notes/content-calendar.md:

| <id> | <Title> | <Target publish date> | BRIEF |

5. Handoff to writer

Write to state/roles/writer/inbox.md:

task: Write outline and script
project: <id>
brief: state/projects/<id>/brief.md
priority: normal
added: <ISO date>

Update state/projects/<id>/status.md → Stage: BRIEF → complete.

Quality Gates

• Brief has all sections populated
• Every idea candidate has a working title, thumbnail concept, four 1–5 scores, and an evidence line
• Selection reason explicitly references scores
• Internal-evidence step ran (or its absence is stated in the brief)
• Angle is stated in one sentence
• Hook is concrete, not vague
• Calendar entry added
• Writer inbox updated

Failure Modes

Failure	Correction
Too many angles, none chosen	Force selection; defer others to future brief
No internal data (new install)	State it in the brief and proceed on external signals — never block
Candidate can't be packaged (no title/thumbnail concept)	Not eligible for selection; bank or drop it
Scores asserted without evidence	Add the search-intent / outlier / internal-data line or lower the score
Hook is generic	Rewrite with specific tension or curiosity gap
No SEO notes	Add at least one primary keyword