S4h

Human

The master orchestration skill. You describe your situation and goal. This skill designs a tailored reasoning workflow — an ordered sequence of skills from across all 23 categories — then executes each step in sequence, feeding what each step reveals into the next. The result is compounding clarity, not a single tool's take.

Your Process

Step 1: Intake

If the user hasn't described their situation, ask:

What are you trying to achieve? Describe the situation and what a good outcome looks like.

Wait for their response before proceeding. Aim to understand:

• The goal: what does success look like?
• The obstacle: what's making this hard?
• The decision point: what are they about to do, decide, or communicate?

If any of these are unclear after reading their description, ask one targeted follow-up question before designing the plan.

Framing check: Confirm the specific situation before continuing. State what you've identified — the goal, the obstacle, and the decision point — in one sentence, then use AskUserQuestion:

• Question: "I'm reading this as: [your one-sentence framing of the goal, obstacle, and what they're about to do or decide]. Is that right?"
• Header: "Framing"
• Options:
  • Yes — proceed — framing is correct
  • Adjust — one element is off; user will correct it before you continue
  • Reframe — different situation than read; incorporate the correction before proceeding

Step 2: Design the Workflow

Design an ordered sequence of 2–5 skills that together address the goal. Each skill in the sequence should serve a distinct function, and the output of each should meaningfully inform the next.

Sequence design principles:

• Start with framing before analysis. If the problem framing might be wrong, begin with assumption-excavator or sensory-structured-observation before applying analytical tools.
• Generate before evaluating. If options are needed, put generation skills (creativity-alternatives, decision-option-mapping, concept-fan) before evaluation skills (decision-criteria-weighting, logic-check, ethics-check).
• Understand people before designing for them. If the goal involves others' reactions, run emotional-motivation-mapping or communication-audience-modeling before communication-objection-mapping or ethics-impact-scan.
• Validate before committing. If the goal ends in a decision or action, close with a stress-test skill (decision-premortem-analysis, constraint-hardness-testing, logic-fixer).
• End with synthesis when the goal is understanding. For exploratory goals, close with a skill that produces a consolidated view (systems-leverage-analysis, narrative-frame-analysis, aesthetic-coherence-check).

Common workflow patterns:

Use these as starting points. Adapt to the specific situation.

Before narrowing: Show the full candidate workflow to the user before committing to a sequence. Use AskUserQuestion:

• Question: "I've drafted a workflow of [N] steps. Before I lock it in, are there any steps you'd flag as especially important, any you'd cut, or any angle I've missed?"
• Header: "Prioritise"
• Options:
  • Proceed with this sequence — the plan looks right
  • Flag one step — user will name a step to prioritise or adjust
  • Add a missing angle — user will describe a lens or skill not yet included

Step 3: Present the Plan

Show the user the designed workflow before executing. Format it as an ordered list with the rationale for each step and the connection to the next:

Here's the plan I'd run for this:

1. **[skill-name]** — [why we start here; what it will surface]
   → feeds into step 2 by [how its output informs the next step]

2. **[skill-name]** — [what it does given step 1's output]
   → feeds into step 3 by [how its output informs the next step]

3. **[skill-name]** — [what it does given steps 1–2's output]
   [last step: what we'll have at the end]

Then ask: "Run this plan, adjust it, or go straight to a specific step?"

Use AskUserQuestion with:

• Question: "Ready to run the plan?"
• Header: "Direction"
• Options:
  • Run the full plan — Execute each step in sequence now
  • Adjust the plan — Tell me what to change before we start
  • Skip to step [N] — Jump directly to a specific step
  • Just run one skill — Pick the single most useful tool and run it now

Step 4: Execute — Sequential with Chaining

Run the first skill immediately. After each skill completes, pause and use AskUserQuestion to offer the next move before continuing.

Before each step (after the first), output a brief handoff:

[skill-name] complete. Key finding: [1–2 sentences on what this step revealed and why it matters].

Then use AskUserQuestion to present what to do next:

• Question: "What's the next move?" (adapt to what was just found — e.g. "We've surfaced 5 assumptions. What now?")
• Header: "Next"
• Options: (build dynamically — 2–3 logical next skills given what the current step revealed, plus a wrap-up option)
  • Label: /[skill-name], Description: [one sentence on why this fits as a follow-on and what it will produce given the current output]
  • (repeat for each next skill, up to 3)
  • Label: "Wrap up", Description: "Synthesise what we have so far and close out"

How to select next skill options:

Pick skills that directly consume the current output. Ask: given what we just learned, what's the most useful next question to answer?

When the user selects a skill, run it on the situation context plus all accumulated output so far. Do not re-explain the situation from scratch — build on what prior steps established.

If the user selects "Wrap up" at any point, jump to Step 5.

Step 5: Synthesize

After all steps complete, output a synthesis section:

What the full sequence revealed:

• List 3–5 findings that emerged from the workflow as a whole — insights that no single step showed on its own.

The key tension or trade-off:

• Name the central conflict the workflow exposed, if any.

Recommended next action:

• One concrete thing to do, decide, or communicate based on everything above.

What to revisit:

• If any assumption or finding in the workflow turned out to be load-bearing but uncertain, flag it here.

Notes

• The goal drives the plan, not the problem type. The same problem ("I need to decide X") could warrant different workflows depending on whether the goal is speed, thoroughness, buy-in, or ethical grounding.
• 2–3 steps is often better than 5. A tight two-skill sequence with strong chaining beats a sprawling five-step plan where the connections are thin.
• Adapt as you go. If step 2 reveals the situation is fundamentally different from what step 1 assumed, redesign the remaining steps rather than running the original plan on stale premises.
• The handoff is the skill. The value of this workflow is what transfers between steps. Never skip the explicit connection from one step to the next.
• If the user just wants one skill, use the routing logic from the original human skill — present 3–4 options, let them pick, run it. Not every situation needs orchestration.