Gptaria publish project

Publish a GPTaria project (Pro order) and pull responses

GPTaria is an informational platform where clients post project briefs and builders respond with working prototypes. This skill is the client "Pro order" flow: post a brief you worked out with your own AI, pay the €10 publication fee, then pull the responses over MCP and build on them here.

Publish only through the GPTaria tools below. Do NOT read source code to do this. Everything you need is in this skill plus the gptaria MCP tools. Never open, clone, search, or read any Git repository, source tree, docs/, or other SKILL.md files to "figure out how publishing works" — you don't need them, and a real user won't have them. Just call create_project.

If the GPTaria tools aren't available in this session (no create_project / list_responses tool), the GPTaria MCP server simply isn't connected. Do not fall back to reading files or hand-crafting API calls from source — instead tell the user to connect it, then retry:

• install the plugin: /plugin marketplace add gptaria-mcp/gptaria-mcp then /plugin install gptaria-mcp, or
• add the server to the MCP config: command npx, args ["-y","gptaria-mcp"], env GPTARIA_API_KEY.

Setup guide: https://github.com/gptaria-mcp/gptaria-mcp

When to use

• The user has an idea (often already developed with an AI in this editor) and wants it posted on GPTaria without re-typing it into the web form.
• The user wants to pull the responses a project received and work on them (compare, pick, implement) in their own assistant.

What the €10 Pro order gives you

• Programmatic publishing — post the brief straight from your AI (no re-typing).
• Auto-translation into every platform language is bundled into the €10 (you are not charged AI tokens for it).
• After it goes live, pull the responses over MCP and work on them here. GPTaria does not analyze responses for you — that happens in your own assistant.

Manual publishing via the website form stays free; the €10 is for the programmatic path (it's also the anti-spam gate on the write API).

Prerequisites

• A GPTaria client account (role: project-starter). Builders cannot publish — a builder key returns 403 not_starter.
• An API key from Settings → API keys, stored as GPTARIA_API_KEY.
• Ability to complete a €10 / 399 UAH checkout in a browser (card / Apple Pay / Google Pay; PayPal via Paddle internationally). You (the user) pay — the assistant never enters payment details.

Tools

With the GPTaria MCP server connected:

• create_project({ title, description, source_language, budget?, category? })
• list_responses({ project_id })
• list_projects()

REST equivalents (header Authorization: Bearer $GPTARIA_API_KEY):

• POST /api/v1/projects
• GET /api/v1/projects/:id/responses
• GET /api/v1/projects

Workflow — publish a project

1. Write the brief in ONE language (the source language), in the user's own words. GPTaria does not generate the brief — your model does.
   • title — 10–200 chars, concrete (what to build).
   • description — 50–2000 chars: the goal, the must-haves, what "done" looks like. Name real tools where relevant (Cursor, Claude, Bolt, Lovable, …).
   • source_language — en | ru | uk | el.
   • budget (optional) — a whole number in the poster's currency (uk → UAH, else EUR). Omit for "negotiable".
   • category (optional) — software | video | audio | image | content | marketing | social | other.
2. Show the user the final title + description and get explicit confirmation — this starts a €10 payment and, once paid, publishes publicly.
3. Call create_project. It returns a draft and a payment.checkout_url:

   { "ok": true, "id": "…", "slug": "…", "status": "draft",
     "payment": { "service": "pro_order", "amount_cents": 39900, "currency": "UAH",
                  "checkout_url": "https://…" } }
   

   The project is not public yet.
4. Give the user the checkout_url to open and pay €10. Do not attempt to pay for them. After payment the project auto-publishes (public, translated) and builders can respond.
5. Keep the id — you need it to pull responses.

Workflow — pull and work on responses

1. list_responses({ project_id }) → { builder, source_message, budget_cents, budget_currency, prototype_url, github_url, is_selected, created_at }[].
2. Work on them here, in the user's own assistant — summarize each approach, compare stacks, help the user pick, and help implement/build from a chosen prototype. (This is the point of the Pro order — the responses come to your editor.)
3. is_selected: true marks a builder the client has already picked.

Guardrails

• Never call create_project without explicit confirmation — it starts a €10 charge and leads to a public post. Show the final brief first.
• Never enter payment details — hand the checkout_url to the user.
• Keep the brief in the user's own words; do not invent requirements.
• Payment for the work is arranged directly with the builder, off-platform — GPTaria never handles money for the work.
• Do not post prohibited content (violence/weapons, drugs, hacking, fraud, CSAM, …).