Set up Econ Review

SkillDev tools

Prepare, verify, repair, or refresh the user-owned Python runtime and local Review Desk used by an installed econ-review plugin. Use when the user installs econ-review from a marketplace, asks to finish setup, reports missing Python packages or Poppler, needs the Review Desk launcher, or wants to check installation readiness without creating duplicate Claude Code or Codex skill copies.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Set up Econ Review skill

What this skill tells your AI

The instructions your AI receives, as published by hanlulong/econ-paper-review-skill in econ-review/skills/econ-review-setup/SKILL.md and read by ahel’s review.

Resolve PLUGIN_ROOT as the parent directory of the skills/ directory that contains this setup skill. Use the standard-library setup tool at PLUGIN_ROOT/scripts/setup_econ_review.py. Do not copy the skill into an agent home: marketplace installation already supplies it.

  1. Find a working Python 3.10 or newer command with venv and pip bootstrapping support already on the machine. Do not install Python, a package manager, a missing python3-venv component, or system software on your own.

  2. Choose global support setup unless the user explicitly requests one-project state. For global setup, run this exact operation first with --dry-run:

    PYTHON PLUGIN_ROOT/scripts/setup_econ_review.py --support-only --global --with-review-desk --dry-run
    

    For project state, replace --global with --local PROJECT_DIRECTORY.

  3. Summarize the dry run in plain language. State that applying it creates or refreshes a private econ-review virtual environment, may download the version-constrained Python packages declared by the plugin, and installs the already bundled, manifest-verified Review Desk. It does not install a second skill copy, use administrator access, or upload a manuscript.

  4. Apply the same command without --dry-run only when the user has explicitly asked to install, finish, repair, or refresh setup, or confirms after seeing the plan. Do not treat a request merely to review a paper as authorization to download packages or change user-level state.

  5. Preserve the setup tool's exit distinction. Exit 0 means the core runtime, PDF ingestion, and requested Review Desk passed. Exit 2 means the managed runtime and Review Desk may be ready but PDF ingestion is still incomplete, normally because Poppler is absent. Do not call that a failed plugin install.

  6. Never install Poppler, Tesseract, TeX, Pandoc, Node.js, an optional PDF backend, or any administrator-managed package automatically. If Poppler is missing, report the setup tool's platform-specific user-level options and ask separately before running a package-manager command.

  7. Verify runtime discovery without changing files:

    PYTHON PLUGIN_ROOT/scripts/setup_econ_review.py --runtime-path
    

    After project-only setup, pass the same project scope instead:

    PYTHON PLUGIN_ROOT/scripts/setup_econ_review.py --runtime-path --local PROJECT_DIRECTORY
    

    This read-only resolver verifies and prints only the managed interpreter. Preserve the setup command's earlier PDF-doctor result and Review Desk launch command or URL when reporting overall readiness. Never expose package-index credentials or environment-secret values in the response.

  8. When the user asks to uninstall support data, show --cleanup-support --global --dry-run (or the matching --local scope) first. Apply the same scope with --confirm-cleanup only after explicit confirmation. Explain that plugin removal alone retains the runtime, descriptor, and Review Desk, while cleanup leaves the plugin and any direct skill copy unchanged.

Signals

GitHub stars
20
Forks
4
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
econ-review-setup
Source
github.com/hanlulong/econ-paper-review-skill