i-have-adhd: readable responses, adapters and evidence
i-have-adhd explained: response structure, not a smarter model
Learn what this output-style skill changes, how its adapters deliver instructions, and where readability claims stop short of evidence.
What you will learn
- A response contract, not another inference engine
- Shorter is useful only when the answer survives
- Keep the product claim smaller than the evidence
Before you start
- Basic Git and command-line knowledge
- Ability to separate observed behavior from an untested claim
Explain the inspected mechanism, plan a reversible trial and interpret evidence without treating style as correctness.
Key takeaways
- Instruction delivery does not increase model knowledge.
- Completeness and honest uncertainty outrank a shorter response.
- Fourteen adapter cases are not a user study.
A response contract, not another inference engine
i-have-adhd is a repository of response-style instructions and assistant integrations. Its central Markdown rules ask the assistant to surface the answer or next action, make progress visible and keep unrelated work out of the immediate reading path. It does not train a model, add factual knowledge or turn an uncertain diagnosis of a software problem into a verified one.
The inspected package and plugin manifests declare version 0.3.0. The repository uses the MIT license, while package.json is private; a package name in that file is not evidence of a published npm package. This series refers to commit 6f1f982, so later installation screens or default behavior may differ from the source discussed here.
Shorter is useful only when the answer survives
Consider a failed authentication test. A useful first line identifies the failing assertion and the next diagnostic check; an unhelpful short answer invents a missing header and tells the reader to change it. Both can look decisive. Evaluate whether the response preserves the evidence and uncertainty, not merely whether a numbered list replaced a paragraph.
The rules include exceptions for explanations, ambiguous requests, destructive work and explicit output contracts. They also say presentation limits must not discard relevant information. Those exceptions matter for learning: a reader asking how a mechanism works needs the causal explanation, even if a reader trying to resume a known procedure benefits from a compact action list.
Keep the product claim smaller than the evidence
The project name and its writing guidance do not establish diagnosis, treatment or benefits for every reader with ADHD. This is a technical review of instruction delivery and output evaluation, not a clinical assessment. Readers without a diagnosis can also prefer structured answers; preference should be asked about rather than inferred from a label.
We executed fourteen isolated cases against the pinned Node hook and OpenCode adapter using synthetic rules. They check loading and configuration behavior, not model comprehension or real assistant compatibility. The upstream evaluation additionally reports a failed release gate. These boundaries are part of the adoption decision, not footnotes to a guaranteed productivity claim.
Decision guide
| Criterion | Option A | Option B |
|---|---|---|
| Best when | You need predictable behavior and easy auditing | You need adaptive optimization and have reliable telemetry |
| Main risk | May leave performance on the table | Can become difficult to explain or debug |
Implementation steps
- 1
Choose one response you currently find difficult to use.
- 2
Identify the information that must survive a rewrite.
- 3
Compare the reading path before comparing length.
Copy-ready example
{
"project": "ayghri/i-have-adhd",
"revision": "6f1f982d0a47c65899af3c5a7450b7098bc65325",
"manifestVersion": "0.3.0",
"newModel": false,
"clinicalOutcomeTested": false
}Frequently asked questions
Does it provide a new model?
No. It supplies instructions and integrations for existing assistants.
Is this a medical tool?
This review establishes only software behavior; it does not establish diagnosis or treatment.
Sources
- i-have-adhd / README.mdSource checked 2026-09-12
- i-have-adhd / package.jsonSource checked 2026-09-12
- i-have-adhd / skills/i-have-adhd/SKILL.mdSource checked 2026-09-12
- i-have-adhd / evals/RESULTS.mdSource checked 2026-09-12