Skip to content
SkillMixed2,799 tokens in use

chatgpt-plugin

Gets a remote MCP server through OpenAI's plugin directory review: auth, package, restricted data, privacy policy, annotations, the submission portal and the rules OpenAI's pages leave out.

680 stars on the repository MIT Licensephuryn/product-engineering · app-stack/skills/chatgpt-plugin
claude plugin install app-stack@product-engineering
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing chatgpt-plugin costs

41tokens idle
2,799tokens in use
15,265tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Mixed

Script-ledMixedModel-led

Scripts do part of the work and the model's judgement does the rest.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: scripts/check_package.py, scripts/test_check_package.py.

Touches

shelldetectedwrites filesdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter8/15
  • Structure19/25
  • Examples6/15
  • Output15/15
  • Size15/15
  • Resources15/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    scripts/test_check_package.py, line 80

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity100
  • Maintenance100
  • Community44
  • Maturity4
  • Momentum40

chatgpt-plugin is for building, submitting or changing a ChatGPT plugin or Apps SDK app. ChatGPT and Codex share OpenAI’s plugin directory, and the skill works with any remote MCP server, in any language, behind any OAuth provider or none.

It lists the decisions to make before the first upload, the rules every submission meets, an eight-step stage order, and a checker script that validates the package folder and its ZIP before each upload.

What You Get

  • A decision table covering sign-in, endpoint origin, tool list, publishing identity, data-residency project and availability
  • Submission rules: authentication declared in mcp.json, discovery open to the portal scanner, _meta accepted on every request, no secrets in tools or replies, explicit tool annotations, listing copy without pricing
  • An eight-step stage order from decision log to answering the review
  • A list of six places where OpenAI’s pages disagree, with the stricter reading to follow
  • check_package.py to validate a package folder and ZIP, with a test file that breaks a valid package one way at a time

Common Use Cases

  • Preparing a first submission of an MCP server to OpenAI’s directory
  • Checking a plugin ZIP before every upload
  • Writing a privacy policy that names every field a reply returns
  • Answering a review decision or changing a listed plugin

What Makes It Unique

It records what the portal and reviewers did in practice, dated, and says that OpenAI’s current pages win wherever they differ from the skill. It ties the ZIP to the product’s test suite so a hand-rebuilt ZIP cannot go stale, and it holds the privacy policy and reply fields equal both ways with a test. It works with any MCP server and does not need the mcp-server skill.

Under The Hood

The skill is markdown, four reference files, a receipts file and two Python scripts. The checker reads a local package folder and ZIP and exits with an error code when a rule fails; the test script builds a sample package and breaks it. The guidance tells the agent to read OpenAI’s developer pages (fetching them as markdown), connect a custom connector to the production server and call each tool, run routing evals against a local server, write package files and a ZIP, and record a review video. Portal steps such as the draft, domain verification and submission are done by the owner in OpenAI’s portal.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers