# ADHD

> Parallel divergent ideation for coding agents — an architectural fix for premature convergence in autoregressive reasoning.

## Docs

- [Introduction](https://adhd.mintlify.app/index.md): ADHD is a skill for coding agents — an architectural fix for premature convergence in autoregressive reasoning.
- [Quickstart](https://adhd.mintlify.app/quickstart.md): Run your first ADHD loop in under two minutes — as an agent skill, from the CLI, or in TypeScript.
- [Installation](https://adhd.mintlify.app/installation.md): Every way to install ADHD — agent skill, CLI, library, Agent SDK, and per-platform paths.
- [How It Works](https://adhd.mintlify.app/concepts/how-it-works.md): The two-phase diverge/focus loop, the isolation invariant, and the architecture underneath.
- [Frames](https://adhd.mintlify.app/concepts/frames.md): The 15 cognitive distortions that drive divergence, how selection works, and how to author your own.
- [ADHD vs CoT & ToT](https://adhd.mintlify.app/concepts/vs-cot-and-tot.md): How ADHD differs structurally from Chain-of-Thought and Tree-of-Thought — the three load-bearing differences.
- [When to Use ADHD](https://adhd.mintlify.app/concepts/when-to-use.md): Use it for open-ended, high-stakes decisions. Skip it for lookups. Plus the honest cost model.
- [The Agent Skill](https://adhd.mintlify.app/usage/skill.md): How the runnable SKILL.md behaves inside your agent — the pre-flight gate, the loop, the output shape, and calibration.
- [CLI Reference](https://adhd.mintlify.app/usage/cli.md): Every adhd CLI flag, with common invocations.
- [Library Reference](https://adhd.mintlify.app/usage/library.md): The adhd-agent TypeScript API — run options, the structured result, and exports.
- [Agent Integration](https://adhd.mintlify.app/usage/agent-integration.md): Call ADHD at decision points inside your own agent loop — the shape that pays the most.
- [Results](https://adhd.mintlify.app/evals/results.md): Head-to-head numbers — ADHD vs a single-shot baseline across six open-ended engineering problems.
- [Methodology](https://adhd.mintlify.app/evals/methodology.md): How the eval suite works and how to run it yourself.
- [Roadmap](https://adhd.mintlify.app/evals/roadmap.md): What's next for ADHD — tracked openly on the issue tracker.
- [Adopters & Ecosystem](https://adhd.mintlify.app/community/adopters.md): Projects that officially ship or integrate ADHD.
- [Contributing](https://adhd.mintlify.app/community/contributing.md): How to contribute a frame, an eval problem, or a loop improvement.
- [How to Make Your AI Think Better](https://adhd.mintlify.app/blog/how-to-make-your-ai-think-better.md): Make AI think better by structuring its reasoning: give it room to think before answering, force multiple independent attempts instead of one, separate generation from evaluation, and match the technique to the problem. Prompting harder helps less than changing the architecture of how the model reas…
- [Why Does AI Always Give You the Same Ideas?](https://adhd.mintlify.app/blog/why-ai-gives-you-the-same-ideas.md): AI gives you the same ideas because of mode collapse and anchoring: alignment training biases models toward typical answers, and generating into one shared context drags every 'alternative' toward the first one. The fix is structural — isolated parallel generation — not a cleverer prompt.
- [Chain-of-Thought vs Tree-of-Thought: Which Should You Use?](https://adhd.mintlify.app/blog/chain-of-thought-vs-tree-of-thought.md): Use Chain-of-Thought for math and multi-step logic, Tree-of-Thought for search and planning, and parallel divergent ideation for open-ended design where many answers are viable. The deciding question: does the problem have one correct answer, a findable path, or a space of options?
- [How to Brainstorm with AI (Without Getting Generic Answers)](https://adhd.mintlify.app/blog/how-to-brainstorm-with-ai.md): To brainstorm well with AI, stop asking one chat for a list. Run the question through several isolated sessions under different vantage points, ban the obvious first answers, then use a separate pass to score and shortlist. Isolation plus a critic beats any single clever prompt.
- [What Is Premature Convergence in LLMs?](https://adhd.mintlify.app/blog/what-is-premature-convergence-in-llms.md): Premature convergence is when an LLM locks onto its first plausible framing and stops exploring, so every subsequent idea is a variation of the first. It's caused by autoregressive anchoring plus alignment-induced mode collapse, and it's fixed architecturally with isolated parallel generation.