| name | nd |
| description | Manually route the current request through the installed neurodivergent-support skill pack. Use only when the user explicitly invokes `/nd` in Claude Code, `$nd` in Codex, selects ND Router from the skills UI, asks to activate ND support for the current task, or requests ND status/off controls. Select the smallest relevant subset of sibling `nd-*` skills and continue useful work instead of merely describing them. |
ND Router
Treat this skill as the manual front door to the pack. The always-on rules remain available; explicit invocation asks for visible routing and immediate application to the current task.
Route
- Use the arguments after the invocation as the task. If none were supplied, use the active task and its latest constraints.
- Preserve a compact ledger: Outcome, Target, Must, Must not, Authority, Done evidence, and Now.
- Select one primary sibling skill and at most two supporting skills. Read their
SKILL.md files before applying them.
- Prefer these routes:
- unclear start or oversized work:
nd-task-initiation or nd-task-decomposition
- explicit finish-all condition:
nd-goal-persistence
- lost context or repeated corrections:
nd-working-memory plus nd-inhibit
- low capacity or overload:
nd-functional-minimums or nd-sensory-load
- explicit frustration or emotional intensity:
nd-emotion-regulation
- repeated task friction:
nd-environment-scaffold
- research spiral, drift, retry loop, or hyperfocus: the matching gate or guard skill
- time, choice, transition, or closure friction: the matching focused skill
- State the selected route in one line, expose the immediate action, and execute it. Do not return a catalog, assessment, or coaching questionnaire.
Use this compact shape when visible scaffolding helps:
ND support: [selected skills]
Outcome: [terminal result]
Now: [one executable action]
Done when: [observable evidence]
Controls
status: show the current ledger, last verified evidence, blocker, and exact next action.off: stop adding optional ND scaffolds for the current task. Preserve safety, explicit constraints, and required agent instructions.- Any other argument: treat it as the task and route it normally.
Guardrails
- Do not infer a diagnosis, capacity, or emotional state from writing style, spelling, silence, or brevity.
- Do not load all sibling skills when one to three cover the immediate barrier.
- Do not let routing replace execution when a safe next action is clear.
- Do not broaden permissions or relax verification because the router was invoked.
