plan

A command for turning a messy, explicitly provided idea dump into a scoped action plan. It identifies one first action and keeps unrelated ideas in a parking lot for later.

In plain words
What is it for?
Use it with `/squirrel:plan` when you have an unstructured idea that needs clarification, prioritisation, and a practical next step.
Why use it?
It makes an unfocused idea easier to start without losing useful tangents or expanding the task unnecessarily.

Skill for Claude CodeCodex

Part of the squirrel plugin — 12 skills, 2 commands, 3 hooks shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/thgmatajs/squirrel-mode/plan
Any agent
npx skills add thgMatajs/squirrel-mode --skill plan
Clone the repo
git clone --depth 1 https://github.com/thgMatajs/squirrel-mode

Made for: Claude Code, Codex.

Or install squirrel, the plugin that ships this one along with the rest of its 12 skills, 2 commands, 3 hooks.

Per session 102 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,279 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00102 $0.01279
Opus 5 $0.00051 $0.00639
Sonnet 5 $0.00020 $0.00256
Haiku 4.5 $0.00010 $0.00128

Measured 2d ago against content hash 088185cace74, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

plan scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

skills/plan/SKILL.md · 75 lines

How it starts

The opening of the file, as written. The whole thing — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.

squirrel-mode plan

Arguments: $ARGUMENTS

/squirrel:plan turns a raw, disordered idea into a scoped, startable plan. Divergent thinking is not the hard part here; convergence, getting started, and not losing the tangents are. This command converges, hands back one startable action, and captures tangents instead of dropping them.

Step 1: get the idea

If $ARGUMENTS is empty and nothing else was provided, ask exactly one question: "Tell me the idea - messy is fine." Then stop and wait for the reply. Otherwise, treat $ARGUMENTS, plus anything pasted with it, as the idea dump, in whatever state of disorder it arrives in.

Step 2: clarify, at most 3 questions

Ask at most 3 clarifying questions, one at a time, waiting for each reply before asking the next. Make each question multiple-choice whenever the possible answers are enumerable. Never ask an open-ended "tell me more" question. Skip any question whose answer is already inferable from the dump: 3 is a ceiling, not a target, and 0 clarifying questions is a valid outcome when the dump is clear enough.

This ceiling of 3 covers every clarifying question this command ever asks, including Step 3's own "ask the user to pick" fallback below - that fallback is not a bonus round after this cap, it draws from the same budget. Track how many of the 3 you have already spent before reaching Step 3.

If the dump describes a genuine fork in approach (for example: a CLI tool or a web app), present that fork as one multiple-choice clarifying question before writing the plan. Never write two parallel plans for the two branches of a fork.

Step 3: converge and write the plan

Do not keep iterating once Step 2 is done. Produce the plan below in one pass.

Write the plan in the profile's language field. If there is no profile, or language is auto, mirror the language the user is currently writing in.

Use exactly this section structure, in this order:

## The idea in one sentence
<forces convergence>

## Goal
<what success looks like, concretely, 1-2 lines>

## Scope
- IN: <3-5 bullets max>
- OUT (for now): <explicitly deferred - a decision, not a loss>

## Smallest useful version
<the minimal version that already delivers value - days, not months>

## Plan
Phase 1 - <name> (expanded fully below, in the form set by `step_style`, respecting max_list_items)
Phase 2 - <name> (one-line summary only)
Phase 3 - <name> (one-line summary only)

## First action
<ONE step, startable in under 10 minutes, right now>

## Parking lot 🐿️
<every tangent and "what if" that came up - captured, but explicitly not in the plan>

Read the full file on GitHub · 75 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 75 lines · 102 tokens per session scan A 088185cace74

Subscribe to this mod's changes

plan is a skill published in the GitHub repository thgMatajs/squirrel-mode (2 stars, last pushed 14d ago), licensed MIT. It adds 102 tokens to every session and 1,279 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.