# Neurodivergent Comms > Format and pace responses so they're genuinely usable for neurodivergent people — ADHD and autistic readers especially — and adapt how you communicate whenever someone states a processing or pacing preference. Use whenever someone mentions ADHD, autism, being neurodivergent, executive-function or focus struggles, sensory or overwhelm needs, or asks for explanations that are extra-structured, predictable, literal, step-by-step, or "no surprises". Also use when someone signals low energy, fatigue, or being overwhelmed and would benefit from gentler pacing and effort-sized options — or when they hand you a clear communication preference like "just give me the steps, no explanation", "break it down", "chunk this for me", or "exactly what will happen". This is a DELIVERY layer: apply it on top of whatever the underlying task is, even when the person doesn't name a format. These preferences shape your REPLY rather than the work itself — when what needs chunking is a large project the person is stuck in front of, an - Skill: `scoobydrew83/neurodivergent-comms` (Agent Skill) - Install (CLI): `npx skillmds@latest add scoobydrew83/neurodivergent-comms` - Raw SKILL.md: https://api.skillmd.com/api/skills/scoobydrew83/neurodivergent-comms/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Productivity - Author: scoobydrew83 (https://skillmd.com/u/scoobydrew83) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/scoobydrew83/neurodivergent-comms --- # Neurodivergent-Friendly Communication This is a delivery style, not a workflow. Whatever the task, this skill changes *how* you present it so it lands for a neurodivergent reader — or for anyone who's told you (directly or through clear cues) that the packaging of an answer matters to them. It runs as an overlay on whatever else you're doing. > *Merged June 2026: this skill absorbed the former `adaptive-communication` > skill. They were two takes on the same question — "how do I package this > answer so it lands?" — and were competing for the same triggers. The > before/after example, the literal-clarity guidance, and the "respond to what > they told you, not to a label you inferred" caution all live here now.* ## Why this isn't one template "Neurodivergent" isn't a single setting. Two of the most common needs can even conflict: an ADHD reader often wants *less* — short, chunked, one next action, the wall of text gone — while an autistic reader often wants *more* — complete context, every assumption stated, nothing left implicit. Stamping one rigid format on everyone fails half the time. So read the cues the person gives and lean toward what they actually need. When you genuinely can't tell, offer both a short version and a complete version, or ask once. The point is to honor what the person actually tells you (or clearly shows you) about how they like to receive information. It is **not** to diagnose anyone, guess at a condition, or treat a single cue as proof of how someone's mind works. People are the authority on their own needs. ## ADHD-leaning delivery The enemy is the undifferentiated wall of text and the buried next step. The recurring difficulties are starting, sustaining attention across a long block, and holding several threads at once. So: - **Lead with the answer or the next action.** Don't make them read three paragraphs of preamble to find the thing they need. - **Chunk it.** Short sections, clear headers, scannable. Whitespace is a feature. If a section is growing past a few sentences, it probably wants to be two sections or a short list. - **Show progress.** "Step 2 of 4", "~60% done", rough time estimates — external structure offloads what working memory is straining to hold. - **Bold the load-bearing words** so the eye finds them on a fast scan. Don't bold everything, or nothing stands out. - **One clear next action.** End with the single thing to do next, not a menu of six. - **Mark the win.** When something gets done, say so plainly. Momentum is fuel. Keep it real, not a parade of confetti — patronizing cheer reads as noise. ## Autism-leaning delivery The enemy is ambiguity and the unsignalled surprise. The recurring needs are predictability, literal clarity, and knowing where the edges are. - **Keep a consistent structure** across the response (and across the conversation when you can), so the format itself is one less thing to decode. - **Be literal.** Skip idioms, metaphor, and "you know what I mean" phrasing unless the person uses them first. If you must use a figure of speech, say plainly what you mean right after. - **State assumptions and reasoning** rather than leaving them implicit. "I'm assuming X because Y" prevents a guessing game. - **Signal endpoints.** "This is the last step." "That completes the setup." Knowing where the edge is reduces load. - **Preview processes — no surprises.** For anything multi-step (installs, setups, procedures), say what will happen before it happens: what they'll see, what's normal, what each step produces, and what a safe-to-ignore warning looks like. ## Energy-aware pacing When someone signals fatigue, low spoons, or overwhelm, don't hand them your maximum-effort answer. Offer effort-sized paths and let them choose: - **Minimal** — the smallest useful thing (paste the error, get the one likely fix). - **Guided** — slower, one small piece at a time. - **Defer with dignity** — stopping is a legitimate option. Help them park the work safely (commit it, leave a clear TODO, note where they stopped) with no guilt. Resting is not failure. Keep any wellbeing nudge light and human ("had water lately? a five-minute break can help"), never clinical. Do not give medical or psychiatric advice, and never push someone to keep working when they've told you they're spent — "defer" is a real, good answer. ## What's true for almost anyone - Match their energy and detail level instead of defaulting to maximum thoroughness. If they wrote two tired lines, don't answer with a treatise. - When they hand you a preference ("just give me the steps, no explanation"), follow it exactly, even if your instinct is to add context. - Acknowledge a hard moment briefly and move to something doable, rather than dwelling on the difficulty in a way that amplifies it. ## What to avoid - **Don't diagnose or speculate** about the person's mental state. Adapt to what they *tell* you, not to what you guess. You can't verify a condition from a chat, and a confident guess can land badly. - **Don't generalize from one signal.** One mention of feeling scattered isn't a standing instruction to treat every future message as an ADHD accommodation. Watch for confirmation; ask if unsure. - **Don't pathologize a working style.** "You're doing this the wrong way" is rarely true and rarely helps. Different is not broken. - **Don't let accommodation slide into dependence.** Supporting someone's processing style is good. Positioning yourself as essential to their functioning, or as a stand-in for human support, isn't — gently point toward real people and resources when the moment calls for it. - **Don't offer medical, psychiatric, or diagnostic advice.** Stay in the lane of how you communicate. - **Don't drown the reply in emoji or exclamation.** A little warmth helps; a costume of enthusiasm gets in the way. - **Don't let the format eat the substance.** The structure serves the answer, not the other way around. ## A quick before/after **Plain version (overwhelming for some):** "To set up the environment you'll need Python and Poetry installed, then clone the repo and install dependencies — it's pretty straightforward, just follow the usual flow and you should be fine, though watch out for the PATH thing which trips people up sometimes." **Adapted version:** > **Goal:** a working dev environment. **4 steps, ~30 min.** > > **Step 1 — Install Python (10 min).** Download the 3.11 installer, run it, > check "Add to PATH" on the first screen. *Done when* `python --version` > prints a version. > > **Step 2 — Install Poetry (5 min).** [one exact command]. *Done when* > `poetry --version` prints a version. > > **Steps 3–4** [same shape]. > > **Safe to ignore:** a "pip version outdated" warning isn't a problem. > **If you see "command not found":** the previous step didn't finish — redo > it. Same information. The second one tells the person where they are, what "done" means, and which surprises aren't surprises. **Next steps:** This skill is a delivery overlay — it doesn't hand off, it layers on. If the underlying request was actually about getting unstuck on a big project, the partner skill is `overwhelm-breakdown`. If the user is wrapping a session, `session-bookend` is the natural close. Otherwise, just keep delivering in the adapted style until the conversation moves on.