uditakhourii/adhd is a skill that enforces divergent ideation across isolated cognitive frames and then prunes traps to deepen the best ideas for coding agents. It addresses premature convergence in autoregressive reasoning caused by anchoring on initial thoughts.
Project overview
It tackles a structural weakness in autoregressive reasoning by spawning N isolated parallel agent calls with zero shared context and mechanically separating generator and critic passes, offering an architecturally enforced approach rather than a prompting shortcut.
Project type
AI Agent · AI Coding · Prompt Engineering
Use cases
Coding & Development
Deployment
Refer to project documentation
License
MIT
Best for
Developers, AI engineers, and researchers who need architecturally enforced divergent ideation before committing to a solution path.
Users who want a separate critic pass that scores ideas for novelty, viability, and fit and flags seductive-but-broken ideas before deepening survivors.
Key capabilities
Spawns N isolated parallel agent calls under different cognitive frames with zero shared context to prevent anchoring during divergence.
Mechanically splits reasoning into a separate critic pass that scores ideas for novelty, viability, and fit, then clusters and deepens the survivors.
Flags seductive-but-broken ideas or traps during the focus phase with one-line reasons.
Limitations and risks
Increases time and output token cost compared to a single-shot baseline.
Requires an underlying LLM to execute the isolated parallel calls, which means availability and cost depend on that model.
Getting started
Install with a single command via `npx skills add UditAkhourii/adhd`, then invoke `/adhd "your problem"` to start. No coding is required.
Evidence and sources
GitHub project description: ADHD — a skill for coding agents. Tree-of-thought with pruning, built on the Claude & Codex Agent SDK.
README: An architectural fix for premature convergence in autoregressive reasoning. Linear Chain-of-Thought anchors on whatever it says first. Tree-of-Thought widens the search but still…
README: 20 traps flagged with one-line reasons — including the cute "stream tokens in reverse" and "patience-token billing" ideas before they cost engineering time.
README: Adopted by [repowire](https://github.com/prassanna-ravishankar/repowire) — the first OSS project to officially ship ADHD.
README: Vendored ADHD as the `think` plugin in their Claude plugin marketplace — wires the divergent-then-converge loop into mstack for architecture, naming, API design, and fuzzy debuggi…