plan-the-big-one
Skill1 starsChanged 19 days agoMITView on GitHub
Plan work too big for one agent session as a map of decisions that any session can pick up. Use when a loose idea or large change has many unknowns, when asked to plan a big feature, migration or refactor, to break it down, to make tickets or a spec, or to ask me the questions first. Plans and asks; does not build.
In this repository17 other files it gives its agents.
- broooooSkill
- did-we-break-itSkill
- finish-the-prSkill
- fix-it-properlySkill
- just-tell-meSkill
- lets-build-pyramidsSkill
- next-time-fasterSkill
- no-added-commentsSkill
- onmywatchSkill
- sharpenSkill
- show-meSkill
- statechartSkill
- talk-is-cheapSkill
- trust-issuesSkill
- trust-me-broSkill
- walk-me-throughSkill
- you-get-meSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
If the work fits in one session, say so and do not make a map.
- Distinctive: line 48
Read first: Read the code, docs and tickets the idea touches.
- Distinctive: line 70
Do not start building until the reader says so.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
plan-the-big-one
Find the way from here to the goal, one decision at a time, and write it down so the next session starts where this one stopped. This skill plans. It does not build.
Idea: $ARGUMENTS. If that is empty, the idea in the conversation.
If the work fits in one session, say so and do not make a map. For one design choice with scale questions, use /lets-build-pyramids.
The map
One file or one issue. Use the tracker the project already uses; if it has none, use docs/plans/<topic>.md, and ask before you create anything in a shared tracker.
# <goal in plain words>
## Destination
What is true when this is finished: a spec, a decision or a shipped change. One or two lines.
## Decided
- <decision>: <the answer in one line>. <where the detail or evidence is>
## Open
- [find out] <question> (blocked by: <other question>)
- [decide] <question>
- [try] <question a rough prototype would settle>
- [do] <manual step a decision waits on>
## Not clear yet
<areas you know are coming but cannot yet ask a sharp question about>
## Out of scope
<work past the destination, and why>- A question goes in Open when you can state it sharply, even if nobody can answer it yet. Otherwise it goes in Not clear yet. Do not chop fog into fake tickets.
- Size each open question to one session.
- A decision lives in one place. The map holds one line and points to the detail.
- Plans and old tickets are research, not requirements. Check them against the code that exists today.
How to work through it
- Read first. Read the code, docs and tickets the idea touches. Write the Destination and show it to the reader. Everything else depends on it.
- Find the facts yourself. Answer every
[find out]question you can from the code, the logs, the docs or the web, with evidence. Never ask the reader for a fact you can look up. Ask in rounds. Ask only the
[decide]questions you can ask now, meaning those whose earlier decisions are settled. Skip any question whose answer depends on another open one. Show each on one concrete example, with the same named people across the round, in plain words:Q1. <the choice, in one plain sentence> Example: <named people and records in one situation> A. <option>: <what happens to them> (recommended: <why, one line>) B. <option>: <what happens to them>"1 A, 2 B" answers a round, and "all as recommended" accepts it.
Each answer changes what is open. Update the map, then ask the next round.
- Challenge your own plan. For each step ask: why do we need this, does it already exist, and could we do less? Remove what does not survive.
- Save before you stop. The map must be enough for a new session with no memory of this one. End by naming the next open question.
When it is done
The map is done when nothing is left to decide before someone can build it. Then:
- Turn it into ordered steps, foundation first, each one checkable on its own.
- Do not start building until the reader says so.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

