{"id":102778,"plugin_id":"plugin_asdk_app_6aeb9369b214819192d2ff3968de771a","kind":"skill","collection_source":"plugin_package","comparison_source":null,"observed_at":"2026-11-18T06:13:27.416Z","digest":"728328f87ff21325a8dd0b3f971159e74bf80b23e7600d33edf2b505246728b3","against":null,"payload":{"description":"Help a user get started with Script.it by inspecting their connected workspace and building, saving, and testing their first automation. Use when the user runs plugin setup, asks to get started with Script.it, or asks to turn the task in this conversation into their first Script.it workflow.","included_files":[{"relative_path":"agents/openai.yaml","size_in_bytes":446}],"name":"get-started","skill_md_contents":"---\nname: get-started\ndescription: Help a user get started with Script.it by inspecting their connected workspace and building, saving, and testing their first automation. Use when the user runs plugin setup, asks to get started with Script.it, or asks to turn the task in this conversation into their first Script.it workflow.\n---\n\n# Get started with Script.it\n\nMove from the user's goal to a verified first result, with a saved workflow or\nagent when that is what the user requested.\nFollow explicit user instructions over these defaults. If the user requests\nonly an explanation, inspection, or a draft, respect that scope.\n\n## Open the workspace\n\n1. Call Script.it's `start_session` before giving generic setup advice. If this\n   conversation already has a usable Script.it session and its complete context,\n   reuse it. Read all returned context, including runtime instructions, skills,\n   workspaces, and connected integrations. Recover any truncated context through\n   documented means before relying on missing details.\n2. Use the returned `session_id` for `run_command` and `read_output`. Follow the\n   live runtime reference for commands and workspace paths. Read relevant skills\n   using the documented commands before building. Keep execution in Script.it.\n3. If authentication is required, use the supported connection flow and resume\n   after it succeeds. Never ask for passwords or tokens in the conversation. If\n   the tools are unavailable, explain the connection needed; do not claim a\n   session opened or substitute a local execution for a Script.it run.\n\n## Choose the first task\n\n- If the conversation already contains a concrete supported task, continue\n  it immediately. Reuse supplied inputs, destinations, preferences, and the\n  requested execution mode. Accept concrete jobs such as scraping a portal,\n  sourcing leads, processing invoices, or building a knowledge agent, alongside\n  \"keep doing this\" and \"when this happens, do that\". Do not require recurrence.\n- If no task is given, ask one short question with three concrete choices\n  grounded in the available capabilities. Include at least one option that\n  needs no additional app connection, such as collecting a public page into a\n  CSV. Use concrete tasks from the use-case guide below; include schedules,\n  event rules, or interactive agents when relevant to the user.\n  Tailor other options to connected apps. Let the user choose or describe\n  their own task before creating a workflow.\n- Ask only for information that blocks progress, such as the source URL, a\n  destination that cannot be inferred, or an ambiguous workspace. Use a returned\n  default workspace when appropriate. Do not ask users to repeat known details.\n- If a needed integration is missing, use the documented connection process.\n  Complete independent work while waiting when useful, and identify the step\n  that remains blocked. Never describe an unconnected app as available.\n\n## Build and verify\n\nOnce the user has chosen a task, follow [Automate a task](../build-automation/SKILL.md)\nusing the existing session and recovered task. It covers all twelve product\nuse cases, one-off execution, saved workflows, interactive agents, schedules,\nand event triggers. Do not repeat discovery or ask the user to restate known inputs.\n\nStarting setup alone does not authorize messages, external writes, or recurring\nruns. Complete actions already requested by the user, including an explicitly\nrequested schedule or event trigger, when the necessary information is available.\n\n## Show the result\n\nLead with what was completed and a small preview of its actual output. Link to\nthe output, saved workflow, or agent using a URL returned by Script.it or a URL\nformat explicitly documented in the live context. If only a session URL is\navailable, label it as the session and identify any saved resource separately.\n\nState whether recurring execution is active, configured but inactive, or not\nconfigured when relevant. Finish with one useful next action, such as changing\nan input or completing an event connection. If the user already requested that action and\nprovided what is needed, complete it instead of offering to do it later.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}