Frames

A frame is a vantage operator: a system-prompt payload that re-poses the entire problem from a different cognitive position. Not a persona, not a domain expert — a deliberate distortion that forces the generator into a corner it would not naturally drift toward. 15 built-in frames ship today, biased toward engineering when codeMode is on (the default). Each is a vantage prompt plus tags.

The frame library

Frame	Vantage prompt	Tags
Hardware engineer	You think in latency, memory layout, and physical constraints. Re-ask this as a hardware/firmware problem. What does the bus topology, cache, timing budget tell you?	code, wild
Regulator / auditor	You audit systems for compliance and failure modes. What must be provable, traceable, or refusable here?	design, general
10-year-old	You are a curious 10-year-old who has never seen software. Describe naive but unencumbered approaches. Ignore convention.	general, wild
Competitor trying to break it	You are a hostile competitor or attacker. Generate approaches that exploit, fail, or sabotage the obvious solution. Then invert into ideas.	code, design
Biology	Transplant a mechanism from biology (immune systems, neural plasticity, cell signaling, evolution, gut flora). Force-fit it onto this engineering problem.	code, wild
Logistics	Steal mechanisms from logistics: queues, batching, just-in-time, hub-and-spoke, returns, last-mile. Apply them literally.	code, design
Game design	Approach this as a game designer. What are the loops, rewards, friction, save-states, speedrun tricks? Treat the user as a player.	design, general
Markets	Treat the problem as a market. Buyers, sellers, market-makers. What does an auction, a futures contract, a clearing house look like here?	design, wild
Inversion	Ask the OPPOSITE question. If the goal is X, brainstorm how to guarantee NOT X. Then negate each answer back.	code, design, general
Extreme: $0 budget, 1 hour	No money, no team, one hour. What is the crudest version that still does the load-bearing thing?	code, general
Extreme: infinite budget, 10 years	Infinite compute, infinite engineers, a decade. What is the maximalist version?	design, wild
Remove the load-bearing assumption	Name the thing everyone treats as fixed (framework, database, request-response model, network). Imagine it is gone. What is possible?	code, design, wild
Speedrunner	You are a speedrunner. Find glitches, skips, out-of-bounds tricks, frame-perfect shortcuts. What is the abusive-but-legal path?	code, wild
Ant colony / swarm	No central planner. Many dumb agents, local rules, pheromone trails. How does the problem solve itself emergently?	code, wild
3am on-call	You are the on-call engineer woken at 3am when this breaks. What design would let you not get paged?	code, design

How frames are selected

codeMode (default true) biases selection toward code and design tags. For code-shaped problems: 4 frames tagged code or design, plus 1 tagged wild. For open product or strategy problems: a mix from all tags.
A wild frame always gets one reserved slot per run so divergence stays weird.
Selection is deterministic per-seed so runs are reproducible — but varies across sessions so the same problem produces different candidate sets when re-run.

Frames vs personas

Frames are not personas. Persona-prompting research (e.g. “you are John, a 34-year-old engineer”) studies simulated identities; some findings report that ordinary attribute-personas outperform curated expert-personas. ADHD frames are neither — they are vantage operators (“you think in latency, memory layout, and physical constraints”) that re-pose the problem, not identities the model role-plays. The mechanisms are different, so persona findings do not map directly onto frame selection. (Tracked in issue #17.)

Authoring your own frame

A frame is ~5 lines in src/frames.ts. Append to the FRAMES array:

{
  id: "supply-chain-attacker",
  label: "Supply-chain attacker",
  prompt:
    "You compromise software by attacking what people TRUST: package managers, build tools, " +
    "CI runners, vendored dependencies, mirror infrastructure. Re-ask this problem as: what " +
    "would I plant, and where, to compromise the result without touching the application code?",
  tags: ["code", "design", "wild"],
},

Field	What
`id`	kebab-case, stable forever — don’t rename
`label`	shown to the user; ~3 words
`prompt`	the vantage prompt, written as an instruction to the generator: “You are X. Re-ask this as Y.”
`tags`	any of `code`, `design`, `general`, `wild`. Tags affect frame selection — code-mode biases toward `code`/`design`; `wild` always has one reserved slot per run.

Quality bar

A good frame passes at least two of:

Distinct vocabulary

Uses concepts no existing frame uses — pheromone trails, futures contracts, frame-perfect skip.

Distinct posture

Adversarial vs constructive vs naive vs maximalist. Not just a different domain saying the same thing.

Reproducible distortion

Consistently surfaces ideas the other frames don’t.

Test it

Run your frame in isolation and check that the ideas are structurally different from what the other frames produce on the same problem:

# tweak src/cli.ts or write a quick scratch script that forces selectFrames
# to return just your new frame, then run:
npm run dev -- "design a write-ahead log under bursty load"

If the ideas are paraphrases of what the “hardware” or “logistics” frame already produces, the frame isn’t earning its slot. Iterate the prompt. Full guide in Contributing.

Getting Started

Core Concepts

Using ADHD

Evals & Results

Community

Blog

The frame library

How frames are selected

Frames vs personas

Authoring your own frame

Quality bar

Distinct vocabulary

Distinct posture

Reproducible distortion

Test it

​The frame library

​How frames are selected

​Frames vs personas

​Authoring your own frame

​Quality bar

Distinct vocabulary

Distinct posture

Reproducible distortion

​Test it

The frame library

How frames are selected

Frames vs personas

Authoring your own frame

Quality bar

Test it