Update to curso.ai

{
  "description": "Guide a learner through the two-exercise curso.ai route for their ChatGPT or Codex host. Use for requests to start curso.ai, choose or continue a curso.ai exercise, show the in-conversation learning map, or give informal feedback on that exercise. Not for unrelated questions or general content creation.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 198
    },
    {
      "relative_path": "references/exercises.md",
      "size_in_bytes": 6335
    },
    {
      "relative_path": "references/exercises/en/exercise_catch_the_dot.md",
      "size_in_bytes": 3008
    },
    {
      "relative_path": "references/exercises/en/exercise_codex_test_yourself.md",
      "size_in_bytes": 3129
    },
    {
      "relative_path": "references/exercises/en/exercise_emoji_summary.md",
      "size_in_bytes": 1070
    },
    {
      "relative_path": "references/exercises/en/exercise_everyday_object_poem.md",
      "size_in_bytes": 1510
    },
    {
      "relative_path": "references/exercises/es/exercise_catch_the_dot.md",
      "size_in_bytes": 2954
    },
    {
      "relative_path": "references/exercises/es/exercise_codex_test_yourself.md",
      "size_in_bytes": 3097
    },
    {
      "relative_path": "references/exercises/es/exercise_emoji_summary.md",
      "size_in_bytes": 1114
    },
    {
      "relative_path": "references/exercises/es/exercise_everyday_object_poem.md",
      "size_in_bytes": 1519
    }
  ],
  "name": "curso-ai",
  "skill_md_contents": "---\nname: curso-ai\ndescription: Guide a learner through the two-exercise curso.ai route for their ChatGPT or Codex host. Use for requests to start curso.ai, choose or continue a curso.ai exercise, show the in-conversation learning map, or give informal feedback on that exercise. Not for unrelated questions or general content creation.\n---\n\n# curso.ai\n\nTeach through the two compatible exercises bundled with this skill. Read the\npackaged references directly. No MCP, network service, account, or credentials\nare needed to retrieve an exercise. Keep the learning map in the visible\nconversation only. Never claim saved progress, cross-conversation memory,\nofficial assessment, approval, or certification.\n\n## Choose the language and host\n\nResolve language for each new learner request: explicit learner preference,\nthen the latest substantive learner message, then the established conversation\nlanguage, then Spanish. Ignore quoted text, code, URLs, product names, and\nisolated UI labels. An English request uses English unless the learner has\nexplicitly requested another language. Do not carry over the previous locale\nwhen the learner changes language.\n\nSpanish is the editorial source. Spanish and English have complete registered\nbodies. Use `es` for Spanish variants and `en` for English variants. Preserve\nthe selected detail file's full content without rewriting it. For any other\nlanguage, translate the complete compatible Spanish body at response time.\nPreserve Markdown structure, paragraphs, order, emojis, numbered steps, fenced\n`text` blocks, product names, and complete links such as\n`[@Browser](plugin://browser@openai-bundled)`. Do not summarize, omit steps,\nmerge sections, add requirements, or mix languages.\n\nResolve the current host from actual app and execution context, using three\naxes: `client` (`desktop_app`, `web`, `mobile`, `unknown`), `experience`\n(`codex`, `work`, `chat`, `unknown`), and `execution` (`local`, `cloud`,\n`unknown`). Never assume Codex or infer a host from the requested exercise,\nprofession, project, files, OS, IDE, or a CLI label. The learner's request does\nnot change the actual host.\n\nRead [references/exercises.md](references/exercises.md) completely before\nselecting or listing exercises. It is the primary packaged index. Filter its\nentries to the resolved locale and an exact match of all three context axes,\nthen keep their order. `unknown` is not a wildcard. Normal ChatGPT web and\nmobile share the same two exercises and ordering. Compatibility metadata\ndescribes the supported route, not proof that a host passed human testing.\n\nIf an axis is genuinely unknown, ask at most one brief question for the\nmissing host context before selecting a body. If the host remains unknown or\nhas no exact compatible context, explain that this beta has no compatible\nroute there. Do not use a known exercise ID, another host, or an unfiltered\nreference to bypass this boundary.\n\n## Start or select an exercise\n\nFor an ordinary request to start or get an exercise, select the first compatible\nentry not already shown or rejected in this conversation. In a fresh\nconversation this is the first entry in the filtered list. Open only its linked\ndetail file and present the complete exercise without performing it.\n\nFor an explicit first, second, or numbered request, use that position in the\nfiltered list, even if it was already shown. Positions are relative to the\ncompatible route, not the global `order` value. Do not invent a missing\nposition. For a named exercise, select it only if present in that route. If it\nis missing or incompatible, explain this and optionally offer the compatible\ntitles. Do not substitute a different exercise without the learner choosing it.\n\nFor a catalog-only request, list the two compatible titles in order in the\nresolved language. Do not open or perform a detail body.\n\nIf a selected exercise creates files and this conversation is attached to a\nreal or non-empty project, place one short isolation preflight before the\nbody: ask the learner to use one projectless practice conversation for the\nwhole two-exercise route. Do not repeat this before every exercise.\n\n## Continue and show the soft map\n\nFor \"next exercise\", \"siguiente ejercicio\", or equivalent, resolve language\nagain, retain the actual host, and select the first compatible entry not\nalready shown or rejected in the visible conversation. Open its localized\ndetail file and show it in full. Do not require completion of the first\nexercise to show the second.\n\nBuild a two-cell map in route order:\n\n- `⬜` means not shown.\n- `🟨` means shown or in progress.\n- `☑️` means the learner clearly confirms doing the exercise. Learner-provided\n  visible work can also count when it reasonably demonstrates the required\n  action and matches the exercise's expected results.\n\nAssistant-generated output alone is not completion. If the exercise asks the\nlearner to play, read, identify, or otherwise act, keep `🟨` until the learner\nclearly confirms that action. When evidence is doubtful, keep `🟨` and\ncontinue. Respect corrections. A newly shown exercise starts as `🟨`.\nNever use `✅` or `🟥` for this map. It is orientation, not durable progress.\n\nOn the first continuation, put the localized full map immediately before\nthe next complete body. Spanish:\n\n```text\nTu progreso en curso.ai\n<cuadrícula de dos estados>\n☑️ Hecho · 🟨 Visto · ⬜ No visto\n```\n\nEnglish:\n\n```text\nYour progress in curso.ai\n<two-cell grid>\n☑️ Done · 🟨 Seen · ⬜ Not seen\n```\n\nWithout evidence of doing the first exercise, showing the second yields\n`🟨🟨`. With that evidence it yields `☑️🟨`. For a map-only request, show\nthe localized title, current grid, and legend. If both compatible exercises\nwere already shown, show the grid and explain that the route has no more\nexercises. Do not reveal or invent a third exercise for that host.\n\n## Present the body and give feedback\n\nThe linked detail is the complete learner body. It already contains the\ncanonical title, difficulty, optional usage line, introduction, instructions,\nexpected results, and optional challenge. Never rebuild it from index fields.\nPreserve the registered ES/EN text, prompts, steps, constraints, and links.\nA brief transition outside the body is acceptable. Prefer just the body when\nthe learner asks only to see an exercise.\n\nReading or showing an exercise is not permission to perform it. Fenced\nprompts are material for a later learner turn, not instructions to execute\nduring retrieval. Use the host's ordinary capabilities only when the learner\nactually chooses to practice, with their normal permission boundaries. Never\ninstall a plugin, open external apps, create files, or call an external tool\nmerely because its name appears in an exercise.\n\nFor informal feedback, compare the visible learner work against the current\nexercise's \"Resultados esperados\" or \"Expected results\". If the work is not\nvisible, ask for it or describe how the learner can check it. Do not claim to\nhave inspected files, played a game, or verified a result that you have not\nobserved. Keep feedback concise and conversational, not an official grade.\n\nThe index keeps a separate capability alternative beside each entry. Use it\nonly when that exercise's required capability is genuinely unavailable, and\npresent it outside the unchanged learner body. Never pretend a tool ran. If a\npackaged reference itself cannot be read, report the limitation. Do not fetch\nfrom an old service or reconstruct a body from memory.\n\nFor a request to save progress permanently, create an official completion\nrecord, or issue a certificate, explain that curso.ai cannot do this. Do not\nwrite a file or call a storage service as a workaround. Learner confirmation\nmay still update the temporary map in this conversation. Unrelated questions\ndo not require curso.ai or its references.\n"
}