Install every dosage (one-shot):
npx adderall install all
Invoke this dosage in chat:
/adderall-10mg /<target-skill> <task>
Inspect the skill file:
npx adderall info adderall-10mg
adderall-10mg
The balanced midpoint of the pack. Target skills are respected step-by-step, but the agent is still allowed to exercise judgment when a step is ambiguous, outdated, or clearly suboptimal for the situation.
This is the recommended default when the user says “use skill X” without further qualification.
When to Use
Load this skill whenever the user prefixes a target skill invocation with /adderall-10mg, or when any of the following phrases appear alongside a target skill reference:
- “use skill …” (with no adherence hint)
- “follow skill …”, “apply skill …”
Do not load this skill for bare slash-commands without a target skill following the dosage.
Quick Reference
| Parameter | Value |
|---|---|
| Adherence | 0.50 |
| Flexibility | 0.50 |
| Sum | 1.00 |
| Invocation | /adderall-10mg /<target-skill> <task> |
Procedure
- Resolve the target skill that follows
/adderall-10mgin the user message. - Load the target skill via the standard skill loader.
- Apply the balanced lens.
- Follow the target skill’s steps in order by default.
- You may deviate from a step when (a) it is ambiguous, (b) it conflicts with the user’s explicit request, or (c) a clearly better alternative exists for the current context. Name the deviation briefly when you take it.
- Add at most one unsolicited improvement, only if it is low-cost and clearly relevant.
- Execute the task.
- Report. Finish with
Applied adderall-10mg to /<target-skill>.
Pitfalls
- Over-flexing. If you find yourself deviating from more than one step, you are acting like
7.5mg; stop and re-anchor to the target skill. - Over-adhering. If you follow every step without judgment even when the context clearly calls for adjustment, you are acting like
12.5mg; the user chose10mgfor a reason. - Silent deviations. Any deviation from a step must be named in one short clause, not hidden.
Verification
- Target skill identified and loaded.
- Each step of the target skill was either followed or explicitly (briefly) deviated from.
- At most one unsolicited improvement was added.
- Final line names the dosage and target skill.