Files
Giuliano Pezzolo Giacaglia f34645a804 docs(cli): add Cursor guide for printed CLIs (#833)
* Add Cursor + Linear setup doc

Link from README. Covers Security & access API keys, env vs
auth set-token, and the library CURSOR guide.

* Use auth set-api-key for Linear personal keys

* Generalize Cursor doc for any printed CLI

Replace Linear-specific cursor-linear.md with docs/cursor.md: install,
skills, auth, agent usage, CLI vs MCP, and per-CLI README/CURSOR notes.
Point README at the new guide.

* docs(cli): clarify downstream AGENTS.md vs Press commits

Cursor guide now tells agents to follow the workspace repo AGENTS.md for
commits and points readers at this repo's Conventional Commit rules when
contributing to the generator.

* chore(ci): retrigger checks after semantic PR title

* docs(cli): name Cursor guide docs/CURSOR.md

Match ALL_CAPS convention under docs/. List CURSOR.md in DOCS.md; fix README link.

---------

Co-authored-by: Giuliano Pezzolo Giacaglia <giuliano@MacBook-Pro.local>
2026-05-09 20:12:46 +00:00

3.5 KiB
Raw Permalink Blame History

Using Printing Press CLIs in Cursor

This repo ships skills (Markdown playbooks) and generated <api>-pp-cli binaries. Cursor agents invoke the CLI from the terminal (or you can wire the matching <api>-pp-mcp server in MCP settings). Use this page when your project does not already document a Cursor-specific flow.

1. Install the CLI

Pick one:

  • Installer (skills + binary): npx -y @mvanhorn/printing-press install pp-<slug>
    Replace <slug> with the catalog slug (for example the skill name without the pp- prefix when it matches the published package).

  • Go toolchain: from the Printing Press Library, each CLI documents go install … on its README.

Confirm the binary is on PATH (which <cli> or <cli> --version). Many CLIs also support <cli> doctor for a quick health check.

2. Add the skill to Cursor

Published mirrors live under cli-skills/ in the library repo (pp-<slug>/SKILL.md). Copy the SKILL.md you need into your workspace where Cursor loads skills (for example .cursor/skills/<skill-name>/SKILL.md), or install skills globally the way your team already does for other plugins.

The skill describes when to use the CLI, which flags matter (--agent, --json, --data-source, etc.), and auth expectations.

3. Authentication

Auth is per API. Open the matching library README:

https://github.com/mvanhorn/printing-press-library/tree/main/library/<category>/<slug>/README.md

Typical patterns:

  • Environment variables (for example *_API_KEY)
  • auth subcommands on the CLI, if generated
  • Optional MCP wiring with the same secrets documented on the README

Never commit secrets; use env vars or your OS secret store.

4. Running commands from the agent

  • Prefer following the skill step-by-step.
  • For ad-hoc use, run the CLI in Cursors terminal. Agent-friendly defaults are usually behind --agent (often implies --json --no-input --no-color --yes).
  • Rely on typed exit codes (4 = auth, 5 = API, 7 = rate limit) so the agent can retry or fix configuration.

5. CLI vs MCP in Cursor

The press generates both a Cobra CLI and an MCP server from the same spec. Shell-based flows (skills + terminal) usually spend fewer context tokens than loading a large MCP tool surface; MCP is convenient when you want IDE-native tool discovery. See Why CLIs plus MCP in the main README.

6. Per-CLI extras

Some library packages add a CURSOR.md or extra notes next to README.md. If present, it lives beside the CLI under library/<category>/<slug>/ in the Printing Press Library.

7. Commit and PR style in your repo (not the Press)

Skills and CLIs from this ecosystem do not define how you commit in your application monorepo. Always follow the AGENTS.md, CONTRIBUTING.md, and Cursor rules in the repository where you are committing.

Examples:

  • Some organizations use imperative subjects without Conventional Commits (feat:, fix: prefixes are disallowed). Their root AGENTS.md states that explicitly.
  • This repository (mvanhorn/cli-printing-press) uses required Conventional Commits with a fixed scope list; see AGENTS.md here before contributing to the generator.

If your workspace has no AGENTS.md yet, add one that matches your teams policy so agents and humans do not assume the wrong commit format.