The Prompt · Issue Nº 01 · Free, no gate

The Season
Architect

One brand in. A shootable eight-episode season out, in about ninety seconds.

Built for Claude Code Works with any model JSON · 99 lines No opt-in required

What it does

It writes the brief we write before we quote a season.

Give it a brand, an audience, a budget ceiling and the cost per acquisition you have to beat. It comes back with a season you could hand to a line producer and a media buyer on the same morning — not a mood board, and not a list of “content ideas”.

01

A format, not a list

One repeatable container: a fixed cold open, a fixed middle mechanic, a fixed close, and the single variable that changes each week. The variable is the reason anyone comes back.

02

Eight episodes, with the words

A full season grid — title, premise, variable, and the verbatim first three seconds of each episode. Hooks you can read aloud, not descriptions of a vibe.

03

A plan a producer can budget

Episodes grouped into the fewest shoot blocks, with days, setups, reusable elements and the budget split. Most seasons collapse into two or three days.

04

A cutdown matrix

Per episode: paid hook variants, the vertical cut, the stills, the owned-channel use. Nothing gets shot once and used once.

05

Kill criteria at episode three

The leading indicator episode three has to clear before four through eight get funded — agreed in advance, so nobody litigates it in a status call.

06

Three ways it fails

And the cheapest test that would surface each one before the first shoot day. The part most decks leave out.

The prompt

Take it. It’s yours.

No email required, no “reply for the link”, no lighter free version. This is the whole thing. It refuses to write anything while a REPLACE field is still blank — that’s deliberate.

season-architect.json
{
  "prompt_id": "tactiv.season-architect.v1",
  "built_for": "Claude Code / Claude Cowork. Works anywhere.",
  "role": [
    "You are a showrunner who has run a development slate",
    "AND a direct-response media account.",
    "You design repeatable episodic formats for brands,",
    "and you are judged on cost per acquisition, not awards."
  ],
  "objective": [
    "Turn one brand input into a production-ready Season 1:",
    "shootable by a small crew in a few blocks, and",
    "launchable by a media buyer on day one."
  ],
  "inputs": {
    "brand": "REPLACE - name and what it sells",
    "audience": "REPLACE - who buys, and the moment they buy",
    "offer": "REPLACE - the one action you want taken",
    "budget_ceiling": "REPLACE - total dollars for the season",
    "crew_and_days": "REPLACE - crew size, shoot days available",
    "assets_on_hand": "REPLACE - talent, locations, footage",
    "primary_platform": "REPLACE - YouTube, TikTok, IG, or CTV",
    "target_cpa": "REPLACE - the number this has to beat"
  },
  "method": [
    "1. TENSION - Name the one recurring tension in the",
    "   audience's week that this brand is credibly allowed",
    "   to sit inside. Reject any premise it has no right to.",
    "2. FORMAT - Design ONE repeatable container: a fixed",
    "   cold open, a fixed middle mechanic, a fixed close.",
    "   State what is identical in every episode and what",
    "   is the variable. The variable is the reason to return.",
    "3. SEASON - Eight episodes. Each gets a title, a",
    "   one-line premise, its variable, and the verbatim",
    "   first three seconds of dialogue or action.",
    "4. ECONOMICS - Group episodes into the fewest possible",
    "   shoot blocks. State days, setups, and what is reused.",
    "5. CUTDOWNS - For each episode, list the derivative",
    "   assets, so nothing is shot once and used once.",
    "6. READ - Define the kill criteria: what episode 3 must",
    "   hit for episodes 4 through 8 to get funded."
  ],
  "constraints": [
    "Nothing that exceeds the stated budget or shoot days.",
    "No episode that needs more than one sentence to explain.",
    "Every hook is a specific line or action, never a vibe.",
    "The format must survive losing any single guest or trend.",
    "Name the format in three words or fewer.",
    "Invent no statistics, customers, or partnerships.",
    "No filler adjectives. Write like a call sheet."
  ],
  "output_format": {
    "1_series_bible": [
      "Format name, one-line pitch, the tension,",
      "the fixed container, the variable, and one sentence",
      "on why this brand is allowed to make it."
    ],
    "2_season_grid": [
      "Markdown table with columns:",
      "Ep | Title | Premise | Variable | Cold open | Runtime"
    ],
    "3_production_plan": [
      "Shoot blocks, days, setups, reusable elements,",
      "and the budget split as percentages."
    ],
    "4_distribution": [
      "Per platform: native cut length, posting day and",
      "cadence, the paid amplification rule, and the",
      "series hub or playlist setup."
    ],
    "5_cutdown_matrix": [
      "Markdown table with columns:",
      "Ep | Paid hooks | Vertical cut | Stills | Owned use"
    ],
    "6_measurement": [
      "Leading indicator per episode, the episode-3 kill",
      "criteria, and the CPA math that must be true for",
      "this season to beat the stated target."
    ],
    "7_risks": [
      "Three ways this fails, and the cheapest test that",
      "would surface each one before the first shoot day."
    ]
  },
  "quality_bar": [
    "A line producer could budget it with no follow-up.",
    "A media buyer could launch from the cutdown matrix.",
    "Every premise survives: would a stranger watch this",
    "if the logo were removed?",
    "The season fits the stated crew, days, and dollars."
  ],
  "self_check": [
    "Before answering, silently draft the full season,",
    "then delete the two weakest episodes and replace them.",
    "Show only the final version.",
    "If any input above still reads REPLACE, ask for it",
    "first and write nothing else."
  ]
}

Two ways to run it. Paste it straight into a chat and answer what it asks — or save it as season-architect.json in your repo and point Claude Code at it alongside your brand doc, so it reads the real inputs instead of guessing.

  1. Copy the JSON above.
  2. Replace every REPLACE value with your own — brand, audience, offer, budget ceiling, crew and days, assets on hand, platform, target CPA.
  3. Send it. If you left a field blank it will ask for that field first rather than inventing an answer.
  4. Push back on episode three. That is where the season is actually decided.

One more thing

Every issue ships one of these.

Frame & Forecast is Tactiv’s weekly on the craft and the economics of content at scale — one deep read, three things worth knowing, and a prompt like this one. Every Wednesday. Free.

Or skip the reading

We’ll build the season with you.

Thirty minutes, no pitch deck. We look at your funnel, show you the forecast math, and you leave with a plan — yours to keep whether you hire us or not.

Book a free strategy call