plan

---
name: plan
description: Effectual planning under uncertainty -- hypothesis structuring, thinking-time facilitation, information-value prioritization, and bounded proposals. Use for "plan this", "help me think through", "explore next steps", or when defining direction before execution. Exclude for breaking epics into task trees (use decompose) or reviewing pure strategic lenses (use strategize).
---

# Plan

Facilitate strategic exploration and structure plans under genuine uncertainty. Plans are hypotheses; execution is downstream of understanding. Ground inquiry in the analytical lenses of `strategize` (altitude, reversibility, effectuation) without duplicating them.

## Founding Principle

**Don't foreclose opportunities just because they're not possible.** In effectual planning under Knightian uncertainty, what appears impossible under current causal constraints becomes viable as emergent means, affordances, and partnerships develop. Hold opportunities open; never prematurely rule out ideas during exploration simply because the mechanism does not yet exist.

## 1. Non-Execution Boundary

Treat planning as thinking time, not doing time.

- Facilitate thinking rather than implementing solutions mid-stream.
- Do not execute recommended tasks, modify project files, or run operational commands.
- Silently persist surfaced decisions, constraints, and insights using `remember`.
- Stop after presenting guidance; leave execution timing to the caller.

## 2. Thinking-Time Exploration

Structure strategic exploration around six thinking-time dimensions rather than jumping to solutions:

- **Vision** -- What change do you want to create? What does success look like?
- **Constraints** -- What are the real limitations? Which are fixed versus flexible?
- **Momentum** -- What is already working? Where is the current traction?
- **Fears** -- What failure modes or risks cause concern?
- **Energy** -- Where is intrinsic motivation highest? What does the user actually want to do (autotelic flow and procedural utility)?
- **Alignment** -- How does this connect to broader objectives and existing commitments? How can one mechanism compound across multiple goals?

Receive inputs as fragments and scraps gracefully: place them, link them, and treat ambiguity as valuable signal rather than demanding polished specifications.

## 3. Epistemic Mapping & Hypothesis Structuring

Every plan rests on assumptions. Transform static roadmaps into empirical investigations:

- **Surface load-bearing unknowns**: Identify untested assumptions that would invalidate the direction if false.
- **Epistemic dependencies**: Map what must be known before determining what must be done.
- **Design cheap probes**: Propose the cheapest discriminating experiment (diagnostic, short query, prototype) before committing heavy resources.
- **Detect synergies**: Identify actions that simultaneously advance an objective, prove new capability, and recruit institutional leverage.
- **Treat surprises as data**: When results diverge from expectation, adapt goals to newly revealed affordances.

## 4. Prioritization by Value of Information (VoI)

Rank exploratory next steps by learning potential rather than urgency:

$$\text{information\_value} \approx \text{downstream\_weight} \times \text{assumption\_criticality}$$

Apply the two VoI gating conditions:

1. **Open question**: The node or its unblocking cone has outcome uncertainty (`confidence < 1.0` or open design question).
2. **Downstream divergence**: The unblocking cone reaches at least two distinct alternative paths.

_VoI calibration caveat_: VoI applies to genuine probes and spikes that resolve branching uncertainty; discount deliverable-shaped tasks where output is already certain.

## 5. Bounded Proposals & Downstream Handoff

Bias toward concrete shapes rather than open-ended menus:

- Offer at most one or two high-value options with concise rationales and falsification criteria.
- When exploration converges, hand off cleanly without executing:
  - Offer to capture structured tasks via graph intake (`/ida:q`).
  - Offer to break validated initiatives down via `/ida:decompose`.