skills/ai-film-pipeline-master/references/phase-01-story-generation/formats-tv.md

# Formats — television

Entry point into the database. Each kit is a ranked short list; open the `cat-*.md` named in the row for the full card.

A format kit answers one question: **given this format's constraints, which techniques earn their place, and why this one rather than the obvious one.** The ranking is the argument.

**Shelf life.** Episode orders, page counts and commissioning fashion move fast; the structural craft does not. Numbers checked 2026-09. Review 2028-09.

---

<!-- GENERATED:toc — do not edit by hand. Generated in the source package -->

## Contents

- [One-Hour Drama](#one-hour-drama)
- [Half-Hour Comedy](#half-hour-comedy)
- [Limited Series](#limited-series)
- [Television Pilot](#television-pilot)
- [Procedural / Case of the Week](#procedural-case-of-the-week)
- [Soap and Continuing Drama](#soap-and-continuing-drama)
- [Coverage](#coverage)
- [Sources](#sources)

<!-- /GENERATED:toc -->

## One-Hour Drama

**Constraints:** 45 to 75 pages depending on where it airs. Broadcast hours are tighter — roughly 40 to 50 pages for about 43 minutes of air, cut into five acts of eight to ten minutes each, and the act breaks are commercial breaks, so they are not negotiable. Streaming hours run 50 to 65 pages with no mandated breaks, and 55 pages is still the centre of the bell curve for a pilot. Cable and streaming shows are frequently written with no act structure marked at all. The writing is a room, not a person: your episode must survive being broken by eight people and produced while episode four is still shooting. Standing sets and a series-regular cast are fixed costs, so the script that uses them is the script that gets made.

**Grammar:** The hour builds meaning by running two or three strands against each other and letting them collide at the act-out. An A-story carries the episode's plot and a B-story carries its argument, and the episode is only good when the two answer the same question from opposite ends. Each act ends on a turn that makes staying through the break the audience's own idea — on broadcast that is a hard structural requirement, on streaming it is the only thing preventing an eight-hour show feeling like a long third act. Character change in an hour drama is measured across a season, not an episode, so each episode moves one degree and pays one debt. The ensemble is a system: a change to one relationship must be visible in two others, or the show reads as a series of separate two-handers. The teaser exists to make a promise the fifth act answers.

**Palette:** 45–75 pages · 4–5 acts (broadcast) · 1 A-story · 1–2 B-stories · 1 act-out per act · 1 degree of season arc · 8–22 episodes

**Rules:** Break the story as A/B/C strands before you write a scene | End every act on a turn, not on a beat of information | Make the B-story argue the opposite side of the A-story's question | Move the season arc exactly one degree per episode and pay one debt | Write into your standing sets; original locations are a negotiation | Open on the strand with the most pressure, not the one that comes first chronologically | Give each series regular one scene per episode where they want something | Put the episode's thesis in the teaser and its answer in the last act | Never resolve two strands in the same scene | Write the act-out first and build the act backwards to it

**Examples:** broadcast five-act hours (*The Good Wife*, *ER*), premium and streaming hours (*The Bear*, *Slow Horses*, *Succession*), the returning-season UK hour

**Avoid:** Act-outs that are merely surprising rather than consequential — the fake cliffhanger that costs nothing is the format's standard failure and audiences detect it within two episodes. The episode where the A-story and B-story are simply two things that happened, with no thematic relation. Wheel-spinning: characters withholding information from each other for no reason except that telling would end the season. The teaser that is a mood piece instead of a promise. And writing a two-hour film split into eight parts, which is now the commonest streaming-drama complaint and shows up as episodes three through six having no shape of their own.

**The kit (ranked):**

1. **The Episode Opening** (Openings) `open.episode_opening`
   An hour drama is entered mid-relationship every week. The opening's job is orientation without recap, and it is the hardest recurring problem in the format.
2. **The Suspended Ending** (Tension) `tens.suspended_ending`
   The act-out. Five times an episode on broadcast, and the reason the format has the shape it has.
3. **The Value Turn** (Scene) `scene.value_turn`
   The tool that survives a writers' room. A scene with a turn can be rewritten by four people and still work; a scene without one cannot be fixed by any of them.
4. **Four-Act Structure** (Structure) `struct.four_act`
   The working skeleton for the modern hour, with broadcast adding the fifth. Start here before deciding to abandon act structure.
5. **The Cold Open** (Openings) `open.cold_open`
   Before titles, before credits, and on broadcast it is doing the work of a trailer for the rest of the hour.
6. **The Ensemble as a System** (Relationships) `rel.ensemble_as_system`
   The difference between an ensemble drama and a set of overlapping two-handers. One change must propagate.
7. **The Serial Clock** (Time) `time.serial_clock`
   One visible deadline running under the season keeps twenty-two episodes from reading as twenty-two incidents.
8. **Thematic Consistency Across a Series** (Theme) `theme.series_consistency`
   The room's shared instrument. It is what lets eight different writers produce episodes that feel like the same show.
9. **The Pair Matrix** (Relationships) `rel.pair_matrix`
   Map every pairing and find the ones you have never played. It is the fastest cure for a season-three staleness problem.
10. **The Micro-Reveal Ladder** (Information) `info.micro_reveal_ladder`
    Every episode must pay something on the season mystery, or the audience learns that waiting costs them nothing.
11. **Rotating the Spotlight** (Relationships) `rel.rotating_spotlight`
    Eight regulars and one hour. The rotation is a structural discipline, not a courtesy to the cast.
12. **The Institutional Antagonist** (Antagonist) `anta.institutional`
    Hospitals, precincts, firms, newsrooms — the hour drama's natural opposition, because it can lose weekly without being defeated.
13. **Interesting Beats Likeable** (Character) `char.likeable_vs_interesting`
    The hour drama lives with these people for years. Likeability is exhausted by episode eight; contradiction is not.
14. **Exposition as Accusation** (Dialogue) `dial.exposition_accusation`
    The weekly re-orientation problem solved without a recap scene: let someone throw the backstory at someone as a charge.
15. **Late Entry** (Scene) `scene.late_entry`
    Forty-three minutes across three strands means roughly fourteen minutes per strand. Every scene starts a page later than drafted.
16. **The Scene Question** (Scene) `scene.scene_question`
    Before it goes in the outline, name what the scene is asking. It is the room's most useful vocabulary.
17. **Who Speaks Last** (Endings) `end.who_speaks_last`
    Act-outs are decided by who gets the last line. Give it to the character who now has the power.
18. **The Season Ending** (Endings) `end.season_ending`
    Resolve, then open. A finale that only opens is a middle chapter, and in the current commissioning climate it may be your last one.
19. **The Ensemble Grid** (Character) `char.ensemble_grid`
    Built once at series bible stage and consulted every week: who overlaps with whom on function, and who is redundant.
20. **The Shifting Coalition** (Relationships) `rel.shifting_coalition`
    The renewable engine. Alliances that re-form under each new pressure generate story without new characters.

---

**The visual kit (ranked):**

1. **Over-the-Shoulder** (Framing) `frame.ots`
   Standing sets, a fixed cast and a weekly schedule mean the two-hander is the format's main event. The OTS pair is what makes it shootable in a morning and readable for six seasons.
2. **Lighting Ratio** (Lighting) `light.ratio`
   The one visual property that can carry a season arc. Establish the show's ratio in the pilot and move it across the run; the audience reads a drift they could not name.
3. **Depth Staging** (Blocking) `block.depth_staging`
   Five acts of eight to ten minutes on standing sets goes stale fast. Staging in depth is how the same room plays differently in episode nine.
4. **Eyeline Match and Eyeline Height** (Angle) `angle.eyeline_match`
   An hour drama cuts between more faces per minute than any other scripted form, and the eyeline is the only thing holding the geography together across that many reverses.
5. **Colour Temperature as Mood** (Colour) `color.temperature`
   Cheapest possible episode identity: a fixed warm-cool map across recurring locations, decided once and free thereafter.
6. **Cut on Action** (Cuts) `cut.on_action`
   The default that makes a five-act hour play as one piece. The alternative, cutting on the line, is what makes broadcast drama feel like broadcast drama.
7. **Medium Close-Up** (Shot size) `size.mcu`
   The format's working size. Wider loses the performance on a television at domestic distance; tighter has nowhere to go at the act break.
8. **Normal Lens (40 to 50mm)** (Lens) `lens.normal_40_50`
   Long glass on a standing set flattens the one thing the set has, which is depth. Reserve the compression for the exteriors.
9. **Motivated Lighting** (Lighting) `light.motivated`
   A recurring interior is believed or it is not, and it is believed because the light comes from the windows and lamps that were there last week.
10. **The 180-Degree Rule** (Angle) `angle.rule_180`
   On a standing set the line is effectively permanent, so it is decided at build and honoured for the run rather than rediscovered each episode.

## Half-Hour Comedy

**Constraints:** About 22 minutes of screen time. Multi-camera runs roughly 45 to 50 pages because the format's page is short and fast; single-camera runs 28 to 35, and 30 pages is the working target. Structure is a cold open, two or three acts and a tag. Multi-camera traditionally holds two acts with a cold open and tag; single-camera takes two to three. Multi-camera is shot in front of an audience on standing sets, so the jokes are tested live and the room rewrites on the night. Single-camera can go anywhere but has no laugh to hide a flat line behind. The A/B/C strand structure is tight enough that the C-story often gets three scenes total.

**Grammar:** Comedy structure is escalation, not plot. The engine is a character whose defining flaw guarantees they will make the wrong choice, placed in a situation where that choice makes things worse, repeatedly, until a reversal. Status is the currency: nearly every laugh in a half hour is somebody's status moving, and a scene without a status differential will not be funny no matter how good the lines are. Jokes come from the situation the structure built, not from wit laid on top — if the joke could move to another scene it is a line, not a beat. The half hour is also where series comedy became emotionally load-bearing: the modern single-camera comedy pays a real feeling in the last two minutes, and it only works because the preceding twenty were funny. The tag is the format's free beat; use it to pay something the episode set up and then walked away from.

**Palette:** 22 minutes · cold open + 2–3 acts + tag · 3 strands · 1 flaw per lead · 4–6 jokes per page (multi-cam) · 1 emotional beat

**Rules:** Build the engine from a flaw that guarantees the wrong choice | Give every scene a status differential before writing a line | Escalate — each complication must be caused by the fix for the last one | Put the funniest word at the end of the line | Never let a joke stop the story to be enjoyed | Plant the callback in the cold open and pay it in the tag | Hold one real emotional beat for the last two minutes and earn it with jokes | Cut any joke a different character could have said | Make the B-story about the same subject as the A-story, played at a lower stake

**Examples:** multi-camera (*The Big Bang Theory*, *Abbott Elementary*'s mockumentary variant sits outside it), single-camera (*Ted Lasso*, *What We Do in the Shadows*, *Hacks*), the British six-episode half-hour

**Avoid:** The joke that stops the story — a great line that costs momentum reads as a sketch inside a series and the audience feels the gear change. Characters becoming their catchphrases by season two, which is what happens when the flaw was never specified precisely enough to generate new situations. The emotional beat that was not paid for: sincerity in the last ninety seconds only lands if the preceding twenty minutes were genuinely funny. Jokes any character could deliver, which is the diagnostic that your voices are not differentiated. And a C-story that is just a runner given scenes it does not need.

**The kit (ranked):**

1. **Comic Character Construction** (Comedy) `comedy.comic_character_engine`
   The half hour's actual engine. Get the flaw specific enough and it generates a hundred episodes; get it vague and it generates a catchphrase.
2. **Status as the Engine** (Comedy) `comedy.status_engine`
   Nearly every laugh in a sitcom is a status move. Set the differential before writing dialogue and the jokes largely appear.
3. **Escalation and the Comic Spiral** (Comedy) `comedy.escalation_spiral`
   The shape of the second act. Each complication caused by the previous fix — the difference between a spiral and a list.
4. **Comic Timing Written, Not Performed** (Comedy) `comedy.timing_on_the_page`
   Multi-camera rewrites on the night, so timing must survive on the page in front of a room full of people at 9pm.
5. **The Funny Word Goes Last** (Comedy) `comedy.funny_word_last`
   The most mechanical and most reliable rule in the format. Applies to nearly every line you will write.
6. **The Topper** (Comedy) `comedy.topper`
   Punchline, then the better punchline. The half hour's economy demands that one joke pays twice.
7. **The Rule of Three** (Comedy) `comedy.rule_of_three`
   The unit of comic structure at every scale here — line, scene and act all run on it.
8. **The Callback Payoff** (Comedy) `comedy.callback_payoff`
   What the tag exists for. Plant in the cold open, pay after the story has finished, and the episode feels designed.
9. **The Straight Man** (Comedy) `comedy.straight_man`
   Without someone playing the scene straight, escalation has no measure and the audience cannot tell how wrong things have got.
10. **The Cold Open** (Openings) `open.cold_open`
    Ninety seconds that must be funny before the audience has committed. It is also where every callback gets planted.
11. **The Incompetence Zone** (Character) `char.incompetence_zone`
    Specify precisely where each character is bad and the writers' room stops inventing new flaws every week.
12. **The Wrong-Sized Reaction** (Comedy) `comedy.wrong_reaction`
    The cheapest laugh available on a standing set, and the one that most reliably reveals character at the same time.
13. **The Running Gag** (Comedy) `comedy.running_gag`
    Series-level comedy. It buys audience loyalty and it also lets you get a laugh in a scene that has no room for a setup.
14. **Chosen Family** (Relationships) `rel.chosen_family`
    The reason the audience returns after the jokes stop being new. Almost every long-running half hour is built on it.
15. **Setup and Punchline** (Comedy) `comedy.setup_punchline`
    The base unit. Most flat sitcom pages are setup problems, not punchline problems.
16. **The Joke That Stops the Story** (Comedy) `comedy.joke_that_stops_story`
    Read as the failure it names. It is the note most half-hour drafts get and it is always about your favourite line.
17. **Emotion Against Type** (Emotion) `emo.emotion_against_type`
    How the final two minutes land: give the feeling to the character the audience has spent twenty minutes laughing at.
18. **Merge Two Scenes** (Scene) `scene.merge_two_scenes`
    Twenty-two minutes across three strands. Merging is usually the only way the C-story gets a shape at all.
19. **Cringe and Social Comedy** (Comedy) `comedy.cringe`
    The single-camera default mode since the mockumentary era. It runs on the audience knowing more than the character.
20. **Comedy of Manners** (Comedy) `comedy.comedy_of_manners`
    Workplace half hours are manners comedies with a photocopier. Rules, violations and the people who enforce them.

---

**The visual kit (ranked):**

1. **Wide Shot / Long Shot** (Shot size) `size.ws`
   Comedy plays wide. The physical joke needs the whole body and the room to be funny, and a comedy cut in close-ups is a comedy that does not trust its own blocking.
2. **The Wide Joke** (Cuts) `cut.wide_joke`
   The rule stated as a cut: hold the wide through the joke and cut only after it lands. Cutting into the joke kills it, every time.
3. **Flat Lighting / Beauty Lighting** (Lighting) `light.flat`
   Multi-camera is lit flat on purpose - four cameras and no shadow to hide a boom. It reads as cheap only when the format is pretending not to be a comedy.
4. **Symmetry** (Composition) `comp.symmetry`
   The frame that makes an absurd situation read as deadpan. It is the single-camera comedy's most reliable device and it costs nothing.
5. **Dinner Table Blocking** (Blocking) `block.dinner_table`
   The ensemble geometry the format returns to weekly. Solve it once for the standing set and every table scene for six seasons is already staged.
6. **Reaction Cut** (Cuts) `cut.reaction`
   Comedy is built in the reaction, not the line. Whose face the cut lands on is the joke's actual punchline.
7. **Cowboy Shot / American Shot** (Shot size) `size.cowboy`
   The size that keeps hands and props in frame while staying close enough to read a face - which is most of what physical comedy needs.
8. **Eye Level** (Angle) `angle.eye_level`
   Comedy is observed, not judged. An angled camera tells the audience how to feel and takes the joke away from them.
9. **Entrance and Exit** (Blocking) `block.entrance_exit`
   The form's engine: who walks in on what. Staged badly it is a door opening; staged well it is the whole second act.
10. **Clear the Frame and Hold** (Blocking) `block.clear_frame_and_hold`
   The button. Everyone leaves, the frame stays a beat too long, and the room finishes the joke.

## Limited Series

**Constraints:** Six to eight episodes is the standard order, occasionally four or ten, running 45 to 60 minutes each — four to eight hours total, released as a block or weekly. Written and often shot as one continuous piece, with a single writer or a very small room, which is why the format attracts feature writers. Closed by definition, which is also the commercial trap: a hit will be asked for a second season and the honest answer is usually an anthology with a new cast. Cast at feature level, budgeted at feature level per hour, and awarded in its own Emmy and WGA categories, which is a real part of why it gets commissioned. A recurring 2026 complaint is that the form has drifted toward minor material at prestige length.

**Grammar:** The limited series is the only television format where the ending can be known and fixed before the first page, and everything it does well comes from that. It builds meaning by controlled disclosure across hours rather than by weekly escalation: the audience's understanding of a single event is rebuilt three or four times, and the last rebuild is the point. Because there is no next season, every setup must be paid, which changes the arithmetic — a thread you cannot pay is a thread you cut, not a thread you defer. Episode shape still matters: each hour needs its own question and its own answer, or the middle episodes become connective tissue. Time is the form's most-used instrument, and multiple viewpoints on the same span is its most-used structure. The finale is judged the way a feature's third act is judged: on whether the cost was real.

**Palette:** 6–8 episodes · 45–60 minutes each · 1 fixed ending · 2–4 viewpoints · 1 event re-read 3 times · 0 deferred setups

**Rules:** Fix the ending before writing episode one and never renegotiate it | Give every episode its own question and answer, not just a position in the arc | Pay every setup — there is no next season to defer to | Assign each episode a dominant viewpoint and let it change what the audience knows | Design the midpoint of the whole series, not of each hour | Re-read the central event at least twice rather than explaining it once | Put the thematic argument in the mouths of two characters who both mean it | Make the finale's cost something established in episode one | Do not write a second season into the ending; if it earns one, make it an anthology

**Examples:** *Chernobyl*, *Mare of Easttown*, *Baby Reindeer*, *Adolescence*, the Emmy and WGA limited-series categories as a working definition of the form

**Avoid:** The eight-hour feature — one story stretched until episodes three to six are corridor, which is the standard criticism of the form and the most common reason a well-reviewed limited series is never rewatched. Prestige length on minor material: the length has to be argued for, not assumed. Withholding the central event past the point where withholding is the only thing happening. An ending that quietly leaves a door open for a season two, which reads as a lack of nerve and damages the one thing this format sells. And a true-story limited series that treats real people as plot functions.

**The kit (ranked):**

1. **The Controlling Idea** (Theme) `theme.controlling_idea`
   Eight hours with one writer and no next season. The controlling idea is the only thing that makes episode five's cut list decidable.
2. **Multi-POV Architecture** (POV) `pov.multi_pov_architecture`
   The form's signature structure: assign viewpoints across episodes so each hour changes what the whole thing means.
3. **The Knowledge Ledger** (Information) `info.knowledge_ledger`
   Track who knows what, in which hour, including the audience. Across eight hours this cannot be held in a writer's head.
4. **The Recontextualising Reveal** (Information) `info.recontextualising_reveal`
   Not a twist — a re-read. The limited series exists to do this two or three times to the same event.
5. **The Midpoint** (Structure) `struct.three_act_midpoint`
   Applied to the whole series, not the hour. Episode four is where a limited series either turns or starts stalling.
6. **Dual Timeline** (Time) `time.dual_timeline`
   The form's most-used instrument, because it lets the ending be present from hour one without being disclosed.
7. **The Cost Paid** (Endings) `end.cost_paid`
   The finale is judged like a feature's third act. A closed form with a costless ending has no reason to have been closed.
8. **The Earned Ending** (Endings) `end.earned_ending`
   Eight hours of investment makes an audience forensic. Anything unpaid will be named in every review.
9. **The POV Handoff** (POV) `pov.handoff`
   How the viewpoint changes between episodes without the audience feeling dropped. The mechanics the architecture needs.
10. **The Midpoint Reveal** (Information) `info.midpoint_reveal`
    In a six-to-eight hour piece, the reveal that changes the question has to arrive by the midpoint or the second half is a wait.
11. **The Same Event Twice** (POV) `pov.same_event_twice`
    The episode structure the form invented for itself. Done once it is a device; done three times it becomes the argument.
12. **Withheld Flashback** (Time) `time.withheld_flashback`
    The honest version of the slow disclosure — withhold the image, not the fact, so waiting is anticipation rather than obstruction.
13. **Multiple Opponents, One Question** (Antagonist) `anta.multiple_one_question`
    Institutions, families and individuals all pressing the same question is what gives a limited series scope without a villain.
14. **The Corruption Arc** (Arc) `arc.corruption`
    A closed form can complete a downward arc that an open-ended series could never finish. It is a large part of what the format is for.
15. **Thematic Consistency Across a Series** (Theme) `theme.series_consistency`
    Here it is not a room-management tool but a compositional one: eight hours that sound like one piece of writing.
16. **The Setup You Forgot to Pay** (Endings) `end.unpaid_setup`
    Run this as an audit before the finale. In an open-ended show an unpaid setup is a future season; here it is a defect.
17. **The Relationship Arc** (Relationships) `rel.relationship_arc`
    Limited series are usually about a pair, not a person. Give the relationship its own want and its own turn.
18. **Real People and the Composite** (Character) `char.real_person_composite`
    Much of the form is true-story adaptation. This is the card that separates responsible composites from convenient ones.
19. **The Chapter Opening** (Openings) `open.chapter_opening`
    Each hour opens as a chapter, not an episode — it can change register, era and viewpoint, and it should.
20. **The Open Ending** (Endings) `end.open_ending`
    Read as the trap it is for this format. An open ending here is usually a season-two hedge wearing an artistic coat.

---

**The visual kit (ranked):**

1. **Lighting Ratio** (Lighting) `light.ratio`
   Four to eight hours is the longest runway any format gives one visual idea. A ratio that opens soft and closes hard across six episodes is an argument nobody has to notice to feel.
2. **Colour Temperature as Mood** (Colour) `color.temperature`
   Mapped per episode rather than per scene. The limited series is the one television form with a designed ending, so the temperature can arrive somewhere.
3. **Scope 2.35 to 2.40:1** (Format) `look.ratio_scope`
   The declaration that this is not a returning show. It costs nothing, it is decided once, and it is the first thing an audience reads before a word is spoken.
4. **Frame Within a Frame** (Composition) `comp.frame_within_frame`
   A closed-ended story about people who are trapped, which is most of them, has its thesis available in the composition of every interior.
5. **Depth Layers: Foreground, Midground, Background** (Composition) `comp.depth_layers`
   Carries the parallel timeline or the second storyline inside the frame, so six hours of intercutting does not become six hours of explaining.
6. **Telephoto Compression** (Lens) `lens.tele_compression`
   Reserved for the surveillance register - somebody watched, somebody followed. In a limited series it can be introduced in episode one and paid in episode six.
7. **Motivated Lighting** (Lighting) `light.motivated`
   Prestige is mostly this: light that has a source in the room. It is also what lets the one unmotivated moment land as a statement.
8. **Close-Up** (Shot size) `size.cu`
   Rationed across the whole order rather than per episode. The face held at the end of hour six is worth what the five hours before it did not spend.
9. **Angle Change as a Power Shift** (Angle) `angle.power_shift`
   Long enough for the audience to learn the hierarchy and to feel it invert. A feature has two hours for this; here there are eight.
10. **Match Cut** (Cuts) `cut.graphic_match`
   The device that joins two timelines without a caption. In a limited series with a dual structure it is load-bearing rather than decorative.

## Television Pilot

**Constraints:** 30 pages for a half hour, 50 to 60 for an hour, with 55 the centre for a drama pilot. It has two audiences and they want different things: a buyer reading it as a business case for a hundred episodes, and a viewer watching it as one story. Read by people who will stop at page ten if nothing has declared itself. Must introduce the world, the engine, the ensemble and the tone while also being a satisfying episode — the only script in television asked to do both. Premise pilots start at the beginning of the story and dominate drama; non-premise pilots show a typical episode and dominate comedy, where the engine matters more than the origin. Sample pilots and commissioned pilots are read differently: the sample is judged on voice, the commissioned one on production reality.

**Grammar:** A pilot is an argument that a series exists. It works by demonstrating the engine — the repeatable pressure that will generate a hundred stories — and then proving it once, live, in front of the reader. Everything else is subordinate: a beautiful pilot with no visible engine gets a meeting and no order. The world's rules are taught by consequence, never by explanation, because the reader is also deciding whether you can write exposition. Characters are introduced by defining action, and the ensemble must be legible as a system of frictions rather than a list of names — the reader is silently asking "what is the argument between these two in episode nine". Tone is set in the first three pages and cannot be corrected later. The last page makes a promise about the series, not just about the next episode.

**Palette:** 30 pages (half hour) / 50–60 (hour) · 4–7 regulars · 1 engine · 1 world rule taught by consequence · 1 complete episode story · 1 series promise

**Rules:** Name the engine in one sentence before writing, then build the pilot to demonstrate it | Declare the show on page one — tone, world, register | Introduce each regular doing something only they would do | Teach the world's rules by consequence, never by explanation | Tell one complete story that also opens the series | Decide premise or non-premise deliberately, not by accident | Show the friction that will still be generating story in episode nine | End on a promise about the series, not a cliffhanger about episode two | Keep the regular count to what a reader can hold — every extra name costs comprehension | Cut any scene that only exists to explain the show

**Examples:** premise drama pilots (*Breaking Bad*, *Lost*), non-premise comedy pilots (*Seinfeld*, *Parks and Recreation* at season two's engine), the sample pilot written purely as a writing sample

**Avoid:** The pilot that is all premise and no engine — a great first episode that nobody can see a season two of, and the most common reason an admired script does not sell. Explaining the show inside the show, usually via a newcomer being given a tour. Too many regulars introduced too fast, which reads as a room the writer cannot hold. The cliffhanger ending, which promises the next episode when the reader needs a promise about the next hundred. And tonal drift across the first act, which tells a buyer the writer does not know what the show is.

**The kit (ranked):**

1. **The Series Opening** (Openings) `open.series_opening`
   A pilot's first page is doing a different job from any other episode's, and most failed pilots fail here by writing episode one's opening instead of the series'.
2. **The Contract Opening** (Openings) `open.contract_opening`
   State what kind of show this is inside three pages. A reader who is still guessing on page ten stops on page ten.
3. **The Controlling Idea** (Theme) `theme.controlling_idea`
   The buyer's real question is what the show is about, not what happens in it. This is the answer, in a sentence.
4. **The Rule of the World** (Openings) `open.rule_of_the_world`
   One rule, taught early and by consequence, sets the terms for every episode that follows. It is the cheapest worldbuilding in the format.
5. **Rule by Demonstration** (Exposition) `expo.rule_by_demonstration`
   A pilot is also an audition in exposition handling. Break the rule in front of the reader instead of describing it.
6. **The Ensemble Grid** (Character) `char.ensemble_grid`
   Built before the outline. Its function here is negative: it shows you which two regulars are the same person and lets you cut one.
7. **The Defining Action Entrance** (Character) `char.defining_action_entrance`
   Seven people, fifty pages. Each one arrives doing something no one else in the cast would do, or they will not be remembered.
8. **The Ensemble as a System** (Relationships) `rel.ensemble_as_system`
   The engine question in relationship form: which frictions will still generate story in episode nine.
9. **Give Someone Something to Want** (Scene) `scene.give_a_want`
   Applied scene by scene, this is what separates a pilot that moves from a pilot that introduces.
10. **The Tailored Opponent** (Antagonist) `anta.tailored_opponent`
    A series-length opposition must be built to press this cast's specific weaknesses, because it has to keep pressing for years.
11. **The Lie in One Line** (Character) `char.lie_in_one_line`
    For the lead, written before page one. It is also, conveniently, most of the series pitch.
12. **Theme Stated Early** (Openings) `open.theme_stated_early`
    In a pilot this is licensed in a way it is not elsewhere — the reader needs the show's question early, and a character can carry it.
13. **The Value Turn** (Scene) `scene.value_turn`
    Pilots are dense with introduction and the scenes that go soft are always the introducing ones. Every scene still turns.
14. **The Casting Line** (Character) `char.casting_line`
    Pilots are read by people casting them in their heads. One line per regular that an actor would want to play.
15. **The Knowledge Ledger** (Information) `info.knowledge_ledger`
    Who knows what at the end of the pilot is the series' opening position. Set it deliberately, not as a residue.
16. **The Deletion Test** (Exposition) `expo.deletion_test`
    Run over every explanation in the pilot. Most survive deletion, and each one that does buys back a page for story.
17. **The Value Landed** (Endings) `end.value_landed`
    The pilot must complete one story. The clearest test is whether a value actually changed by the last page.
18. **The Suspended Ending** (Tension) `tens.suspended_ending`
    Use it for the series promise, not for a next-episode hook. The distinction is most of the difference between a good pilot ending and a cheap one.
19. **Tonal Drift** (Voice) `voice.tonal_drift`
    Read as the failure it names. It is the single most damaging note a pilot can get, because it reads as the writer not knowing the show.
20. **The Voice Bible** (Voice) `voice.voice_bible`
    Written during the pilot, used by the room for years. Building it now is also how you find out whether your regulars are distinguishable.

---

**The visual kit (ranked):**

1. **Defining Entrance** (Blocking) `block.entrance_exit`
   A pilot introduces eight people who must be distinguishable in episode nine. Each arrives doing something only they would do, and the entrance is staged before it is written.
2. **Wide Shot / Long Shot** (Shot size) `size.ws`
   The world, once, early. A pilot that never shows where it is has asked a buyer to imagine the show's biggest asset.
3. **Lighting Ratio** (Lighting) `light.ratio`
   The pilot sets the show's ratio for its entire run, which is why it is a format decision rather than an episode one.
4. **Colour Temperature as Mood** (Colour) `color.temperature`
   Establish the recurring locations' temperatures here. Every later episode inherits them free, and a pilot that leaves them undecided costs the series its identity.
5. **Over-the-Shoulder** (Framing) `frame.ots`
   Proves the show is shootable at pace. A pilot full of unrepeatable setups is a pilot that describes a series nobody can afford to make.
6. **Frame Within a Frame** (Composition) `comp.frame_within_frame`
   The pilot's one chance to state the show's visual thesis in a single image that the marketing will then use for three years.
7. **Medium Close-Up** (Shot size) `size.mcu`
   The size the series will live in. Establishing it in the pilot is what makes a later departure mean something.
8. **Motivated Lighting** (Lighting) `light.motivated`
   Sets the rule the standing sets get built to. Decided late, it becomes a lighting problem repeated for a hundred episodes.
9. **The 180-Degree Rule** (Angle) `angle.rule_180`
   Fix the line on every recurring set here, and every subsequent director inherits a geography instead of inventing one.
10. **Eye Level** (Angle) `angle.eye_level`
   A pilot is arguing that these people are worth returning to. Angles that editorialise in episode one leave nothing for the season.

## Procedural / Case of the Week

**Constraints:** 42 to 44 minutes of air in five broadcast acts, 22 episodes a season, shot on standing sets with a fixed cast and a weekly guest cast. A problem is presented, investigated and closed inside the hour. Episodes must be watchable out of order and in syndication, which means no episode can depend on another — the serialised layer has to be removable without breaking the case. Guest cast, not regulars, carry the emotional variable, so the weekly guest is where the writing actually lives. Budget and schedule are the tightest in scripted television: eight-day episodes, overlapping units, scripts delivered while two others shoot.

**Grammar:** The procedural makes meaning by contrast between a constant and a variable. The regulars are the constant — they do not change, and their stability is the product, not a limitation — and the weekly case is the variable that tests them without altering them. Structure is a question with a wrong first answer: the case appears to be one thing for two acts, is revealed as another at the midpoint, and closes on a third understanding. Fair play is the audience's contract: every piece of information needed to solve it is on screen before the solution, and violating that is the one unforgivable sin in the genre. The A-story is the case, the B-story is a regular's small personal pressure, and their only relationship is thematic. Competence is the pleasure being sold — watching people who are good at something do it well — so scenes exist to display expertise as much as to advance plot.

**Palette:** 42–44 minutes · 5 acts · 1 case · 1 wrong first answer · 1 midpoint turn · 1 B-story · 2–4 guest characters · 22 episodes

**Rules:** Plant every clue before the solution — the fair-play contract is not optional | Give the case a wrong first answer that holds for two acts | Make the B-story rhyme with the case thematically, never intersect with it plot-wise | Keep the regulars constant and put the change in the guest cast | Write the guest character a want, not a function | Open on the crime or the problem, before the regulars | Show competence in action at least twice an episode | End each act on a development in the case, not on a personal beat | Make the episode work for a viewer who has never seen the show and for one who has seen two hundred

**Examples:** *Law & Order* and its franchise, *House*, *Elementary*, *Criminal Minds*, the medical and legal case-of-the-week hour, the fall 2026 network schedule which is still largely built on them

**Avoid:** The clue withheld until the solution, which is the genre's cardinal sin and the reason a procedural loses an audience permanently rather than gradually. The guest character who is a plot function with a name — the format's most-fixable and least-fixed weakness. A B-story that resolves the case, which destroys the episode's shape and the format's whole contrast. Regulars who change substantially, which breaks the syndication promise the show is sold on. And the wrong first answer that nobody in the episode ever actually believed, which makes the midpoint turn feel administrative.

**The kit (ranked):**

1. **The Fair-Play Contract** (Information) `info.fair_play_contract`
   The genre's foundational agreement. Break it once and the audience stops playing along, which is the entire pleasure they came for.
2. **The Flat Arc** (Arc) `arc.flat`
   The regulars do not change and that is the product. Understanding this as a design, not a constraint, is what separates procedural writers from drama writers doing a procedural badly.
3. **Competence First** (Character) `char.competence_first`
   The thing being sold every week is skilled people working. Establish it before anything personal.
4. **Goal, Conflict, Disaster** (Scene) `scene.goal_conflict_disaster`
   Twenty-two episodes a year at eight days each. This is the scene shape that can be outlined fast and rewritten under schedule.
5. **Procedural Exposition** (Exposition) `expo.procedural_exposition`
   How the technical material gets delivered without stopping the hour. It is the genre's defining craft problem.
6. **The Cold-Open Reveal** (Information) `info.cold_open_reveal`
   The body, the symptom, the filing. The case must be on screen before the regulars, every week, without exception.
7. **The Red Herring** (Information) `info.red_herring`
   The wrong first answer that holds for two acts, built to be believable rather than to be wrong.
8. **Delayed Decoding** (Information) `info.delayed_decoding`
   The audience sees the clue and does not understand it. It is how you satisfy fair play and still land the solution.
9. **The Competent-Character Walkthrough** (Exposition) `expo.competent_walkthrough`
   Expertise delivered as action rather than lecture — the difference between a doctor explaining and a doctor working.
10. **The Professional Pair** (Relationships) `rel.professional_pair`
    The renewable relationship engine. Partners who cannot leave and disagree about method generate story for two hundred episodes.
11. **The Lesser Guilt** (Information) `info.lesser_guilt`
    The suspect is guilty of something else. The oldest and still the most efficient way to make a red herring pay.
12. **Reversal of Expectation** (Scene) `scene.reversal_of_expectation`
    The midpoint turn, at scene scale. A procedural is a chain of small ones, which is why it does not feel like a formula from inside.
13. **The Incompetence Zone** (Character) `char.incompetence_zone`
    Constant regulars still need friction. Specifying exactly what each is bad at is what generates it without character change.
14. **The Episode Opening** (Openings) `open.episode_opening`
    Must orient a first-time syndication viewer and a two-hundred-episode viewer in the same thirty seconds.
15. **The Moral Test** (Theme) `theme.moral_test`
    What stops case-of-the-week being a puzzle with no stake: the case asks the regular something they have to answer.
16. **The Exception** (Character) `char.the_exception`
    The one case a year that gets under a regular's skin. It is how a flat-arc show earns its rare, tiny movements.
17. **The Double-Meaning Clue** (Information) `info.double_meaning_clue`
    A detail that reads innocently first and damningly second. The most re-watchable thing a procedural can plant.
18. **Chekhov's Gun** (Information) `info.chekhovs_gun`
    Nothing appears in a procedural scene by accident; the audience is actively cataloguing. Place deliberately or be misread.
19. **The Serial Clock** (Time) `time.serial_clock`
    The removable serialised layer. One season-long pressure that can be lifted out for syndication without breaking any hour.
20. **The Institutional Antagonist** (Antagonist) `anta.institutional`
    Bureaucracy, jurisdiction, the budget, the chief. The opposition that can push back weekly and never has to be defeated.

---

**The visual kit (ranked):**

1. **Over-the-Shoulder** (Framing) `frame.ots`
   Twenty-two episodes on standing sets with a weekly guest cast: the OTS pair is the unit the whole schedule is built on.
2. **Interrogation Geometry** (Blocking) `block.interrogation_geometry`
   The format's one recurring set-piece. Solved once as geometry, it plays for twenty-two weeks and stays legible with a different guest each time.
3. **Flat Lighting / Beauty Lighting** (Lighting) `light.flat`
   The procedural's honest default. It is fast, it is consistent across directors, and the format's pleasure is the puzzle rather than the atmosphere.
4. **Insert Shot** (Cuts) `cut.insert`
   The evidence. A procedural is a sequence of objects looked at closely, and the insert is where the case is actually made.
5. **Top-Down on Hands** (Angle) `angle.hands_top_down`
   The lab and the desk, weekly. It is the one angle the format can repeat without fatigue because the thing under the hands changes every episode.
6. **Medium Close-Up** (Shot size) `size.mcu`
   The interview size. Close enough for a lie, wide enough for two people, and repeatable by any director on the rota.
7. **Eyeline Match and Eyeline Height** (Angle) `angle.eyeline_match`
   With a new guest cast every week, the eyeline is the only continuity tool that does not depend on knowing the actors.
8. **Reaction Cut** (Cuts) `cut.reaction`
   The regulars' faces are the series; the guest's face is the episode. Which one the cut lands on is the format's whole editorial position.
9. **Motivated Lighting** (Lighting) `light.motivated`
   Keeps twenty-two episodes shot by eight directors looking like one show, because the sources are fixed in the set rather than chosen per shot.
10. **The 180-Degree Rule** (Angle) `angle.rule_180`
   Non-negotiable here. The audience's attention is on the puzzle, and a crossed line spends it on the geometry instead.

## Soap and Continuing Drama

**Constraints:** Continuous production at a pace no other scripted format attempts — episodes airing daily or several times a week, with writing, shooting and post overlapping permanently. The writing is an assembly line by design: a head writer sets long-term story projections running up to two years ahead and maps the week scene by scene; six to ten associate writers work beneath them; breakdown writers produce the day's narrative and script writers dialogue it, typically one writer per day of the week, with a script editor making sure the days track. Deadlines cannot slip, because there is no gap in the schedule to absorb a slip. Standing sets, a large permanent cast and cast availability drive story as much as story does. No ending exists.

**Grammar:** A continuing drama builds meaning by accumulation and by the audience's memory. Nothing resolves permanently; every resolution is the setup for the next configuration, and the pleasure is watching a fixed community rearrange itself under new pressure forever. Information is the currency — who knows what, who is about to find out, who will be in the room when they do — and the form runs almost entirely on the gap between what the audience knows and what the characters do. Secrets are assets with a maturity date: planted, held for months, and paid when the calendar needs them. Every scene ends on a turn or a look, because the form has more scene endings per hour than any other. Because viewers drop in and out, information is repeated constantly, which makes the usual prohibition on characters telling each other what they already know the one rule this format has to break on purpose and with craft.

**Palette:** 3–5 strands per episode · 6–10 writers · 1 head writer projecting 2 years · scenes of 2–4 minutes · 1 secret per strand · 0 permanent resolutions

**Rules:** Project the long story two years out, then break the week scene by scene | Run three to five strands per episode and never resolve two in the same week | Plant a secret with a maturity date, not a vague intention | End every scene on a turn or on a look | Keep the audience ahead of the characters, always | Rotate the spotlight so every regular has a live want at all times | Re-deliver known information as accusation, threat or plea — never as recap | Configure and reconfigure the same relationships instead of adding people | Write for a viewer who missed three weeks and one who has watched for twenty years

**Examples:** *Coronation Street*, *EastEnders*, US daytime (*General Hospital*, *The Young and the Restless*), telenovela and continuing-drama formats worldwide

**Avoid:** Genuine recap dialogue — characters flatly reminding each other of events, which is the format's structural temptation and is solved by re-delivering the information as a weapon, not by deleting it. The secret with no maturity date, which sits unpaid until the audience stops caring, and then paying it costs more than it earns. Resolving a triangle completely, which removes an engine and leaves nothing in its place. Adding characters instead of reconfiguring existing ones, which is what happens when the pair matrix has not been consulted. And treating a cast departure as a plot problem rather than as the free structural event it is.

**The kit (ranked):**

1. **Dramatic Irony** (Information) `info.dramatic_irony`
   The format's central engine. The gap between what the audience knows and what the character knows is what makes a two-minute kitchen scene unbearable.
2. **The Serial Clock** (Time) `time.serial_clock`
   A wedding, a trial, a due date, a return. One visible date gives a formless continuous story its spine for eight weeks at a time.
3. **The Suspended Ending** (Tension) `tens.suspended_ending`
   Every episode, and in the UK format every commercial break inside it. The most frequently applied card in the whole database, per hour.
4. **The Secret** (Character) `char.the_secret`
   A soap secret is an asset with a maturity date. Plant it, hold it, pay it when the calendar needs it, and never let it expire quietly.
5. **Information Asymmetry Between Characters** (Information) `info.character_asymmetry`
   Map who knows what across the whole cast; the map, not the plot, is what generates next month's scenes.
6. **The Shifting Coalition** (Relationships) `rel.shifting_coalition`
   The renewable engine of a fixed community. Re-form the alliances and you get new story without new cast.
7. **The Triangle** (Relationships) `rel.triangle`
   The form's oldest structure, and it works because it is never closed — it is rotated.
8. **Information the Listener Already Has** (Dialogue) `dial.known_information`
   Everywhere else this card names a failure. Here it names the format's compulsory problem, and the card is the only honest guide to breaking it on purpose.
9. **Rotating the Spotlight** (Relationships) `rel.rotating_spotlight`
   Thirty regulars, five strands a week, and actors under contract. The rotation is a production instrument as much as a dramatic one.
10. **The Public Scene** (Scene) `scene.public_scene`
    The pub, the ward, the square. Private conflict staged where the community can hear is what generates next week's strands for free.
11. **The Group with a Shared Secret** (Relationships) `rel.group_secret`
    A secret held by three people has a half-life. Its collapse is a month of story that writes itself.
12. **The Interrupted Scene** (Scene) `scene.interrupted_scene`
    The confession stopped by a door. The form's most-used tool for converting a scene's payment into another month of credit.
13. **The Debt Between Them** (Relationships) `rel.the_debt`
    Obligation is what keeps people in a community they would sensibly leave, which is the format's basic plausibility problem.
14. **The Secret the Reader Holds** (Information) `info.audience_only_secret`
    The audience-only secret is how the format converts a quiet episode into suspense without spending any plot.
15. **The Status Flip** (Character) `char.status_flip`
    A continuing drama runs on reversal of standing inside a fixed community — the returning exile, the exposed pillar.
16. **The Deferred Payoff** (Emotion) `emo.deferred_payoff`
    Only this format can hold a plant for two years and pay it to an audience that still remembers. It is the form's one unmatched advantage.
17. **Who Speaks Last** (Endings) `end.who_speaks_last`
    With more scene endings per hour than any other format, the last-line decision is made hundreds of times a month. Make it consciously.
18. **The Long Marriage on the Page** (Relationships) `rel.marriage_on_the_page`
    The long-running couple is the audience's anchor, and writing them as settled rather than as a solved problem is the hardest thing in the format.
19. **The Corruption Arc** (Arc) `arc.corruption`
    Continuing drama can run a slow fall across three years, which no other format can sustain — and it doubles as an exit for a departing actor.
20. **Enmity of Kin** (Situations) `sit.polti_13_enmity_of_kin`
    The situation the form returns to more than any other. Worth reading as a generator rather than a cliché, because it is going to be used either way.

---

**The visual kit (ranked):**

1. **Over-the-Shoulder** (Framing) `frame.ots`
   Continuous production at a pace no other scripted form attempts. The OTS pair is the only two-hander coverage that can be lit once and shot all day.
2. **Flat Lighting / Beauty Lighting** (Lighting) `light.flat`
   Not a compromise here but the format's engineering: several crews, overlapping schedules, and any actor able to walk onto any set at any hour.
3. **Medium Close-Up** (Shot size) `size.mcu`
   The size the form is written in. Soap is faces receiving information, and this is the size that carries it at speed.
4. **Clear the Frame and Hold** (Blocking) `block.clear_frame_and_hold`
   The cliffhanger, staged. The episode ends on a held face or a held room, and the format makes that decision several times a week.
5. **Close-Up** (Shot size) `size.cu`
   The act-out. Spent at the end of every part, which is exactly why the rest of the scene must not be shot in it.
6. **Entrance and Exit** (Blocking) `block.entrance_exit`
   The form's motor: who comes through the door of a room where something has just been said.
7. **Eyeline Match and Eyeline Height** (Angle) `angle.eyeline_match`
   Scenes shot out of order, weeks apart, by different crews. The eyeline is the only thing that survives that.
8. **Motivated Lighting** (Lighting) `light.motivated`
   Standing sets lit from their own practicals stay continuous across a production that never stops to relight.
9. **Depth Staging** (Blocking) `block.depth_staging`
   The cheapest variety available: the same four rooms, played from a different depth, read as different scenes.
10. **Eye Level** (Angle) `angle.eye_level`
   The register the audience has been watching for decades. Departures are for the once-a-year episode, and they work because everything else is level.

## Coverage

Six kits: One-Hour Drama, Half-Hour Comedy, Limited Series, Television Pilot, Procedural / Case of the Week, Soap and Continuing Drama. Film formats live in `formats-film.md`; social and short-form in `formats-social.md`.

## Sources

Consulted 2026-09-17.

- [How Long Is a TV Show Script — StudioBinder](https://www.studiobinder.com/blog/how-long-is-a-tv-show-script/)
- [How Long Is a TV Show Script: Essential Guide — Filmmaking Lifestyle](https://filmlifestyle.com/how-long-is-a-tv-show-script/)
- [How to Structure and Format Your Television Scripts — The Script Lab](https://thescriptlab.com/features/screenwriting-101/8930-how-to-structure-and-format-your-television-scripts/)
- [Television Script Format — Movie Outline](https://www.movieoutline.com/articles/television-script-format.html)
- [Anatomy of a Contest-Winning 1-Hour TV Pilot — Shore Scripts](https://www.shorescripts.com/anatomy-of-a-contest-winning-1-hour-tv-pilot/)
- [The Evolution of TV Scripts in the Streaming Era — Paste](https://www.pastemagazine.com/tv/streaming-tv/script-advice)
- [How Many Acts Does a TV Show Have? — Screenwriting.io](https://screenwriting.io/how-many-acts-does-a-tv-show-have/)
- [Structuring Your 30 Minute TV Comedy Script — BCG Pro](https://www.comedy.co.uk/pro/inside_track/structuring-your-30-minute-script/)
- [Acts — Celtx Help Center](https://support.celtx.com/hc/en-us/articles/6764181913111-Acts)
- [Cold Open — Wikipedia](https://en.wikipedia.org/wiki/Cold_open)
- [What Is a Limited Series? Definition & Key Examples — StudioBinder](https://www.studiobinder.com/blog/what-is-a-limited-series/)
- [Outstanding Limited or Anthology Series 2026 — Television Academy](https://www.televisionacademy.com/awards/nominees-winners/2026/outstanding-limited-or-anthology-series)
- [The Limited Series Are Down Bad — Infralog](https://www.infralog.in/the-limited-series-are-down-bad/)
- [3 Types of TV Pilots Writers Need to Choose From — The Script Lab](https://thescriptlab.com/features/screenwriting-101/9666-3-types-of-tv-pilots-writers-need-to-choose-from/)
- [What Is a Pilot Episode and How to Write One — Final Draft](https://www.finaldraft.com/blog/what-is-a-pilot-episode)
- [Premise Pilot — Erik Bork](https://www.flyingwrestler.com/2014/04/premise-pilot/)
- [Style and Structure in Procedural Television — Script Magazine](https://scriptmag.com/style-and-structure-in-procedural-television)
- [The 2025-26 Network TV Season Can Reshape The Single Protagonist Procedural — Screen Rant](https://screenrant.com/network-tv-2025-2026-season-single-protagonist-procedural-drama-trend/)
- [Head Writer — Wikipedia](https://en.wikipedia.org/wiki/Head_writer)
- [My Life Writing for the Soaps — NYCityWoman](https://www.nycitywoman.com/my-life-writing-soaps/)