The 5-Step Analog Horror Writing Framework (Works With Any AI Tool)

Analog horror lives or dies on controlled information. The dread comes from what a reader is not told yet — a signal that should not exist, an archive that remembers too much, an entity whose rules you can almost work out. The hard part is not inventing the eerie idea; it is keeping that idea coherent for eighty thousand words.

This is a tool-agnostic framework you can run with any AI assistant, a notebook, or both. It is the manual version of the staged process inside the Dead Signal Book Builder — useful whether or not you buy a dedicated skill.

Step 1 — Define the signal and its rules

Start with the source of unease and pin down exactly what it can and cannot do. In analog horror the “signal” is whatever carries the threat: a broadcast, a tape, a camera feed, a found document.

Write three things before anything else:

• What it does — the specific fear mechanism. “A radio broadcast that describes your room before it happens” is specific. “A scary signal” is not.
• What it cannot do — the limit. A threat with no limit cannot build tension, because the reader stops believing any escape.
• What it costs — the price of engaging with it. If keying the microphone gives the entity one reply, that cost must stay fixed for the whole book.

Fixing the cost early is what stops later scenes from cheating their way out of danger.

Step 2 — Build the archive logic

Analog horror is usually assembled evidence: logs, transcripts, timestamps, maintenance tickets, forum posts. Decide how the archive exists and why it survived, because that frame answers the reader’s first sceptical question — “why is any of this recorded?”

Ask: who kept these records, why do they still exist, and who could have altered them? The tampering question matters most. An archive nobody could edit is a safe archive; an archive one character could have quietly changed is a frightening one.

Step 3 — Design the reveal ladder

A reveal ladder is an ordered list of what the reader learns and when. Each rung should recontextualise the one before it. Write it out explicitly — do not trust yourself to pace reveals by instinct across a whole manuscript.

A simple ladder for an abandoned-station premise might read:

1. The station was demolished years ago.
2. The voice on the broadcast knows the narrator’s name.
3. The previous operator filed the same shift logs, word for word, decades earlier.

Notice that each rung is a fact, not a vibe, and each one makes the previous rung worse. If a rung does not change the meaning of an earlier one, it is probably a scene, not a reveal — move it off the ladder.

Step 4 — Architect chapters and scenes before drafting prose

Most AI writing tools can produce a convincing paragraph and then lose the thread of a book. Fix that by giving every chapter a card before you draft a single line of prose. Each card needs four fields:

• Goal — what the point-of-view character is trying to do.
• Turn — the moment the situation changes.
• Cost — what it takes from them (tie this back to Step 1’s fixed cost).
• Hook — the unresolved question that pulls the reader into the next chapter.

Draft the scene from the card, not from a blank page. This is also where an AI assistant is genuinely useful: a tight card is a far better prompt than “write a scary chapter,” and it keeps the generated prose inside your established rules.

Step 5 — Run an adversarial continuity pass

Before the manuscript reaches a reader, attack it. Go through the draft specifically hunting for broken canon: names that drift, timeline contradictions, a threat that quietly gained a new power, a reveal that landed before its setup.

An AI assistant can help here — paste the story bible and ask it to flag contradictions against the established rules — but treat the output as a second pair of eyes, never a guarantee. AI misses subtle contradictions and sometimes flags an intentional ambiguity as an error. The final judgment is yours.

Keep a running canon tracker as you go, so the adversarial pass is checking against a single source of truth rather than your memory.

Putting it together

These five steps are the same backbone, whether you run them by hand, with your own prompts, or inside a dedicated tool:

1. Define the signal and its rules.
2. Build the archive logic.
3. Design the reveal ladder.
4. Architect chapters and scenes.
5. Run an adversarial continuity pass.

If you would rather follow a guided version of this process with the worksheets already built, our Dead Signal Book Builder review walks through its 11-stage Claude workflow and how it maps onto these same five decisions. Buyers through our link also qualify for a free Bonus Vault with a story-bible worksheet, reveal-ladder planner and 30 scene starters.