Planban commands in Codex and Claude Code

Which commands does Planban add?

Other MCP hosts get no commands. Their agents use the MCP tools directly.

How do I type them in Codex and in Claude Code?

In Codex, type /planban and choose a Planban action from the / menu. The menu shows each command by its display name: PB, Planban, Planban Create, Planban Help, Planban Tutorial and Planban Feedback. You can also mention the plugin, as in @planban Open my Planban board.

In Claude Code, type /planban: to list the Planban commands, then pick one. Claude Code always shows the planban: prefix, because it names commands after the plugin that provides them.

What does each command do?

• pb opens the best matching board at once, with no explanation first. It opens the current project's board when the project is set up for Planban. Otherwise it opens the only registered board, or the all-boards selector when there are several.
• planban opens the same board as pb for a plain open request. It is also the entry point for any other Planban work: reading and changing Items, Specs, Plans and Status. For that work it loads the full Planban protocol first. If you describe a Planban problem, it hands over to planban-feedback.
• planban-create turns rough intent into Planban structure: a new board or project set-up, one new Item, or several Items and Groups drawn from notes, issues or exported plans. See Set up Planban for a project.
• planban-help returns a short guide: the commands for your host, a few getting-started steps, the Item and Group vocabulary, and suggested prompts.
• planban-tutorial starts the local app, creates or reuses the Planban Demo board and opens the interactive tutorial.
• planban-feedback investigates a bug, request or rough edge using the conversation you have already had, searches existing issues, and drafts the best route. Nothing is posted until you approve it. See Troubleshooting Planban.

Can I use plain prompts instead of commands?

Yes. Plain prompts work in any host when they name Planban clearly. These are the prompts Planban suggests:

• Open my Planban board.
• Show all my Planban boards.
• Summarize this project's Planban roadmap state.
• Start work on the Planban roadmap item called <title or id>.
• Create Planban roadmap items from these notes: <notes>.
• Send Planban feedback: <feedback>.
• Check whether Planban has updates.

After /pb or /planban opens a board, a short follow-up such as "start this card" or "do the next thing" is treated as Planban work. If it is unclear which Item you mean, the agent asks.

What does the reply contain after a board opens?

Every reply after a board opens includes the exact board URL as a clickable link, even when the board also appeared beside your session. When the host showed the board, the reply reads:

Planban is open: [Open the verified board](URL)

When the host could not show it, or has no browser beside the session, the reply reads as follows, with at most one short reason:

Planban is running: [Open the verified board](URL)

The tutorial replies the same way, with Planban tutorial is open or Planban tutorial is running and a link to the tutorial. A reply that says only "Board opened" without the link is never correct.

Which command can run without being typed?

In Codex, planban and planban-feedback allow implicit invocation. Codex can use them when your prompt clearly asks for Planban work or describes a Planban problem, without you typing a command. pb, planban-create, planban-help and planban-tutorial run only when you choose them.

In Claude Code, Claude can also use a Planban skill when your request matches its description, such as asking to open your Planban board.