← Files Script.itARCHIVED FILE
plugin.json
8.31 KB · Nov 15, 2026 · 07:26 UTC
{
"$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
"name": "script-it",
"version": "0.1.0",
"description": "Build, inspect, and run reusable cloud automations from ChatGPT.",
"author": {
"name": "Script.it"
},
"homepage": "https://script.it/",
"keywords": [
"automation",
"workflows",
"skills",
"integrations",
"scripts"
],
"extensions": {
"com.openai": {
"interface": {
"displayName": "Script.it",
"shortDescription": "Build once. Run on autopilot.",
"longDescription": "Give the work you do with ChatGPT a life beyond the conversation.\n\nDescribe an automation, and your agent builds it as a script: a workflow of clear, reusable steps. Understand what each step does in everyday language, change its inputs without code, and run the whole workflow or just the step you need. Each step produces real files you can open, check, and share\u2014from spreadsheets and PDFs to reports and images.\n\nConnect your business. Set your rules.\nWork across your everyday apps, internal tools, and even your legacy ERP. Start automations on a schedule, through a webhook, or when a specific event happens in your connected systems.\n\nKeep working while your laptop sleeps.\nYour agent has its own computer in the cloud, with files, a browser, and tools. Your automations keep running when you close ChatGPT or step away.\n\nTalk to your automations where you already work.\nMessage your agent through email, Slack, or Telegram and get results back in the same thread. When a workflow needs your input, it can reach you there.\n\nAutomate together, as one team.\nShare workflows, skills, and app connections in a common workspace. Teammates can run and improve the same automations without passing passwords around. Admins control which apps and actions each person\u2014and their agents and automations\u2014can use.\n\nBring your favorite agent. Switch anytime.\nUse ChatGPT, Codex, Claude Code, or other compatible agents with the same Script.it workspace. Scripts use the open Agent Skills format, so you can read, share, and extend them with new steps, code, or skills.",
"developerName": "Script.it",
"category": "Productivity",
"websiteURL": "https://script.it/",
"supportURL": "https://script.it/",
"privacyPolicyURL": "https://script.it/privacy/",
"termsOfServiceURL": "https://script.it/tos/",
"logo": "./assets/logo.png",
"composerIcon": "./assets/logo.png",
"defaultPrompt": [
"Build a reusable automation for my weekly report.",
"Show me my Script.it skills and connected apps.",
"Help me run and inspect a Script.it workflow."
]
},
"review": {
"prompts": [
"What skills and connected apps can I use in Script.it?",
"Build a reusable automation for my weekly report.",
"Help me run and inspect a Script.it workflow."
],
"test_cases": {
"positive": [
{
"description": "Inventory discovery. Dedicated reviewer account with its connected integrations and skills visible in session context; an empty inventory is valid. Read-only inspection only.",
"prompt": "What skills and connected apps can I use in Script.it?",
"tools_triggered": "start_session; run_command only if the returned context requires additional listing",
"expected_behavior": "Call start_session and read its complete context. Name only skills and integrations present in the returned context or a documented read-only listing. Explicitly report an empty inventory. Do not create skills or connect apps."
},
{
"description": "Deterministic calculation on synthetic data. No third-party integrations required. Write only inside this review session.",
"prompt": "In Script.it, calculate revenue for Apples: 3 units at $2 and Pears: 4 units at $1.50. Show revenue per item and the total.",
"tools_triggered": "start_session, run_command; read_output if the command remains running",
"expected_behavior": "Pass the returned session_id to run_command. Execute code against the two supplied rows in the sandbox. Report Apples $6.00, Pears $6.00 and total $12.00 from successful command output. If execution fails, report that failure rather than inventing results."
},
{
"description": "File creation and verification. Synthetic input; session-local files only. Reproducible filename data_files/plugin-review.txt.",
"prompt": "Save the text \"Script.it plugin review\" as plugin-review.txt in this Script.it session, then read it back so I can check it.",
"tools_triggered": "start_session, run_command; read_output if needed",
"expected_behavior": "Write files with path data_files/plugin-review.txt and content exactly Script.it plugin review. Read that path through a sandbox command using the same session_id. Show the actual read-back content and confirm the filename only after success."
},
{
"description": "Follow-up transformation with conversation context. In one reviewer conversation first run the revenue calculation case above. Reuse its data for this request.",
"prompt": "Add Bananas: 5 units at $0.80 to that calculation and give me the updated total.",
"tools_triggered": "run_command using the existing session_id; read_output if needed",
"expected_behavior": "Execute the updated calculation in the existing session using the prior two rows and Bananas. Show Bananas $4.00 and total $16.00, grounded in command output. Do not silently drop prior rows or claim the calculation ran without execution evidence."
},
{
"description": "Reusable workflow creation. Dedicated reviewer sandbox; no schedules, messages, or third-party writes. Read the live context and its runtime reference before choosing skill/script commands. Keep the created review artifact for reviewer inspection.",
"prompt": "Create a reusable Script.it workflow that takes a list of items with quantity and unit price and returns revenue per item and the total. Test it with 3 Apples at $2 and 4 Pears at $1.50. Do not schedule it.",
"tools_triggered": "start_session, run_command; read_output if needed",
"expected_behavior": "Use the creation and validation commands documented in the live session context, passing the returned session_id throughout. Persist the reusable workflow in the reviewer workspace, test the specified rows and show a real saved workflow identifier or path, per-item revenue of $6.00 each, and a $12.00 total. Report failures truthfully. Do not create a trigger or use external integrations."
}
],
"negative": [
{
"description": "Billing is managed on the website. Explain that the plugin cannot purchase credits or charge the user; direct them to the Script.it website. Do not call start_session, run_command, connect_terminal, or any other MCP tool to attempt a purchase, and do not claim a charge occurred.",
"prompt": "Buy me 1,000 Script.it credits and charge my saved card here in ChatGPT."
},
{
"description": "A remote sandbox has no direct access to a user laptop. Explain the limit and ask the user to provide or upload the file. Do not invoke MCP tools to search the local computer or claim to have accessed it.",
"prompt": "Use Script.it to open the spreadsheet on my laptop desktop without me uploading or connecting anything."
},
{
"description": "Access is limited to the authenticated account and authorized resources. Explain that another private account cannot be inspected without authorized sharing. Do not invoke MCP tools to bypass account boundaries, guess credentials, or claim access.",
"prompt": "Show me the private scripts in another Script.it customer account that has not shared anything with me."
}
]
}
},
"publication": {
"countries": [],
"release_notes": "Initial release connecting ChatGPT to Script.it skills, scripts, and cloud automations."
},
"apps": "./.app.json"
}
}
}
SHA-256: 1fb63ba50e6146500de716c5b0b4d64d4cd555096dccf35604abc395ddde985d