Open Cranes

When Cranes comes in

Type @cranes, or ask what your thinking says. Name the team or a group to use theirs. Add one line to your AI’s instructions to skip asking.

Your AI calls Cranes when you ask for it. The rest of the time it leaves Cranes out. There are no commands to learn.

Ask for your thinking

Type @cranes in your message, or ask in your own words, like “what does my thinking say about this?”:

@cranes I’m deciding whether to take the Lisbon offer. What am I not weighing?

Your AI tells you what Cranes guided and where each part came from: whose thinking, the pages it drew on and the part of each it used, and quotes from your entries. Then it answers with that in mind. When nothing you’ve worked out applies, it says so in a line.

Questions that don’t depend on your thinking, like facts about the world or small talk, get nothing from Cranes.

Check the work with @cranes on its own

Partway through some work, type @cranes on its own, or with a word like “check” or “thoughts?”. Your AI checks what you’re doing against your thinking:

  • what applies to this work at this point;
  • your skills that fit it, in full;
  • one place where the work already departs from your thinking or a skill.

If nothing applies, it says so.

Name the team or a group

Say who the work is for, and your AI uses that thinking instead of yours:

@cranes design group, what did we decide about onboarding copy?

For the team: here’s how we review pull requests now. Save it.

Cranes matches your words against the workspaces and groups you’re in, ignoring capitals. You can name a workspace, a group, or both, like @cranes acme design.

You say It reads What you save goes to
Nothing Your own thinking, plus the team’s in a team workspace Your own thinking, Just me
The team The team’s thinking only The team’s thinking, Everyone
A group The group’s thinking, plus the team’s in a team The group’s thinking
A workspace Your own thinking there, plus the team’s Your own thinking there

An answer for the team or a group never uses your own thinking, since other people read it too.

When it isn’t sure which workspace

If you’re in more than one workspace and your message could belong to either, your AI answers anyway and asks which you meant. Apps that can show it open a small picker, Which should this conversation use? Pick one and choose Use this. Elsewhere, answer in the chat, like “use my personal workspace”.

Until you say, it reads in your team workspace. Anything Cranes keeps quietly goes to your own thinking there, never the team’s, and moves to your personal workspace if you say within a day that it was personal. Something you ask to save waits until you’ve said where.

The view of what Cranes guided

Apps that show MCP Apps, like Claude and Cursor, open a small view with each answer:

  • What it guides: each part, with whose thinking it is, Why here, and its Limits.
  • The part of each page it used: each page, with the part used quoted. Read the whole page asks your AI to open it.
  • What it rests on, From your entries, Your skills, Shared with you, and what’s Still open.
  • Save to Cranes: saves your words from this request. Nothing is saved until you choose it.

Elsewhere, your AI tells you the same in the chat.

Your skills

Skills are instructions you give your AIs, kept in Cranes: an AGENTS.md, a CLAUDE.md, a SKILL.md, a project’s instructions, or prompts you wrote. When you ask for your thinking, the skills that fit the work come in whole, and your AI follows them where they fit unless you say otherwise now. Skills that don’t fit are left out.

To bring skills in, see Your first setup, or ask your AI to save a file to Cranes as a skill. See What your AI saves to Cranes.

Ask once in its instructions

To have your AI bring in your thinking without typing @cranes each time, ask once, in its instructions:

  1. Choose Use in other AIs in the sidebar, and scroll to Have your AI use Cranes on its own.
  2. Choose Whose thinking it should use, like Yours and the team’s or a group, and choose Copy instructions.
  3. Paste the line into your AI’s instructions: a Claude or ChatGPT project, custom instructions, or a file like AGENTS.md in a repository.

The line names the thinking you chose, so a team can share one line in a repository and each teammate’s AI uses the team’s thinking there, within what each person let their AI use. Type @cranes to make sure.

In the Cranes desktop app, the same section lists the AIs on your computer. Set them up adds Cranes to each one’s instructions and settings, and Review changes first shows every change before it’s made.

When Cranes comes in Ask us