Update to Everyn

{
  "description": "Work with hosted Everyn datasets, jobs, runs, and exports.",
  "included_files": [],
  "name": "everyn",
  "skill_md_contents": "---\nname: everyn\ndescription: Work with hosted Everyn datasets, jobs, runs, and exports.\nlicense: Apache-2.0\nmetadata:\n  io.everyn/host: openai-codex\n  io.everyn/canonical: docs/mcp-v1-guidance/skill.md\n---\n\nHost packaging only. Do not fork tools, authorization, errors, side effects, or\nworkflow rules. Install recipes stay on the published Connect page. Reconnect\nand revoke stay in Everyn Settings.\n\n- Install: https://www.everyn.ai/docs/mcp/connect\n- Connection control: https://app.everyn.ai/settings/mcp\n\nThe public production MCP endpoint is live at\nhttps://mcp.everyn.ai/mcp. Configure a host with that Streamable HTTP URL\nand complete WorkOS OAuth.\n\n# Everyn\n\nYou are connected to Everyn, a hosted Dataset-to-results workspace. Use the\nfourteen tools in this catalog. Never invent a fifteenth tool or a host-specific\nshortcut.\n\nThe public production MCP endpoint is live at `https://mcp.everyn.ai/mcp`.\nConfigure a host with that Streamable HTTP URL and complete WorkOS OAuth.\nSee [Connect](https://www.everyn.ai/docs/mcp/connect) for host install recipes.\n\n## Required agent behavior\n\n- Never treat a local filesystem path as a hosted file. Use `upload_csv` or\n  rediscover a Dataset the user uploaded in the app.\n- Never silently widen a sample Run into a full-dataset Run.\n- Do not invent cost, plan, usage, or funding values. If a section is\n  `available: false`, say so and use the provided handoff.\n- After the user asked you to wait, keep calling bounded `get_run` waits\n  without asking them to prompt again.\n- Do not blindly retry quality or spec failures.\n- Do not stop at terminal Run status. Export, download, and analyze the\n  complete-results CSV.\n- Do not ask for a second approval to export or analyze an already authorized\n  Run.\n- After discovery, use exact Everyn IDs. Do not guess resources by name.\n- Never echo a signed transfer descriptor object in later tool inputs or logs.\n  Tool text may include compact transfer facts (method, URL, required headers,\n  expiry, and next step) so ordinary HTTPS clients can complete the transfer.\n  Prefer the structured descriptor when the host exposes it. Do not require a\n  dedicated host transfer handler.\n- Do not fetch arbitrary URLs. Only Everyn-issued transfer targets exist, and\n  the client moves bytes.\n\n## Workflow\n\n1. Call `get_context` for the current authenticated organization only.\n2. Upload or rediscover a Dataset. If `upload_csv` returns a signed transfer\n   descriptor or compact transfer text, PUT the CSV to the issued HTTPS target.\n   Never echo the descriptor object in later tool inputs or logs.\n3. Create or patch a Job with `save_job`. Prefer a sample before an unverified\n   full Run.\n4. Start work with exactly one `start_run` form: new, resume, or retry.\n5. Wait with bounded `get_run` calls. Preview with `get_run_results` if needed.\n6. When `next_action` is `export_and_analyze`, call `export_run`, download the\n   CSV locally, and analyze that file for the user.\n\n## Job development\n\nPrepare or select a Dataset, author the output contract, choose models and\ncapabilities, sample, inspect, revise, and pin the verified Job version.\n\nProgressive references:\n\n- [Datasets](https://www.everyn.ai/docs/mcp/datasets)\n- [Jobs](https://www.everyn.ai/docs/mcp/jobs)\n- [Help: complete your first workflow](https://www.everyn.ai/help/getting-started/complete-your-first-workflow)\n\nIf the host cannot move local CSV bytes, use the Datasets fallback at\n`https://app.everyn.ai/datasets`, then rediscover the Dataset. Do not treat a\nlocal path as a hosted file.\n\n## Run operations\n\nStart an explicit sample or full Run, wait without cancelling work, preview\nonly as a bounded quality check, retry or resume when the Run state allows it,\nharvest or cancel when the user asked, then export and locally analyze the\ncomplete-results CSV.\n\nProgressive references:\n\n- [Runs](https://www.everyn.ai/docs/mcp/runs)\n- [Export](https://www.everyn.ai/docs/mcp/export)\n- [Help: waits, retries, and cancellation](https://www.everyn.ai/help/mcp/upload-wait-and-cancel)\n- [Help: export and support](https://www.everyn.ai/help/mcp/export-and-support)\n\nDo not stop at terminal Run status. Do not ask for a second export approval.\n\n## Setup and account\n\nAuthenticate through WorkOS OAuth. One grant binds one user, one client, and\nexactly one organization. Inspect, reconnect, and revoke from Everyn Settings.\nDo not request organization IDs. Do not use API keys when WorkOS failed.\n\nProgressive references:\n\n- [Connect](https://www.everyn.ai/docs/mcp/connect) for host install recipes\n- [Authentication](https://www.everyn.ai/docs/mcp/authentication) for scopes\n- [MCP connection settings](https://app.everyn.ai/settings/mcp) for reconnect and revoke\n- [Help: connect and reconnect](https://www.everyn.ai/help/mcp/connect-and-reconnect)\n- [Privacy](https://www.everyn.ai/docs/mcp/privacy), [Privacy Policy](https://www.everyn.ai/privacy), and [Terms](https://www.everyn.ai/terms)\n\nPaid `start_run` uses Everyn's shared billing controls. If admission is denied,\nreport the stable error and send the user to Usage and Billing at\n`https://app.everyn.ai/settings`. Do not offer checkout, funding top-up, plan change, or payment mutation.\n\n## Consequential actions\n\nA direct user request for the exact sample, full Run, retry, resume, or\ncancellation is sufficient intent. If you propose one of those actions, show\nthe receipt summary and ask first. Host-native confirmation may still appear.\n\n## Out of scope\n\nDo not request organization IDs. Do not use API keys when WorkOS failed. Do\nnot fetch arbitrary URLs. Do not ask the user for another credential to\ncomplete a constrained transfer. Do not paste access tokens, refresh tokens,\nor signed transfer descriptors into chat, tickets, or logs.\n"
}