Update to Sovrnti

{
  "description": "Turn Sovrnti benchmark results into a short decision brief for an executive, a board, a security lead or a procurement team, with the sources stated. Use it when the user asks for a summary, a memo, a board paper or a recommendation based on Sovrnti results. An example is \"write a one-page brief for our CIO on collaboration suites in Germany\".",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 402
    }
  ],
  "name": "decision-brief",
  "skill_md_contents": "---\nname: decision-brief\ndescription: Turn Sovrnti benchmark results into a short decision brief for an executive, a board, a security lead or a procurement team, with the sources stated. Use it when the user asks for a summary, a memo, a board paper or a recommendation based on Sovrnti results. An example is \"write a one-page brief for our CIO on collaboration suites in Germany\".\n---\n\n# Write a decision brief\n\nExplicit user instructions take priority over this workflow, including on length and format.\n\n## Inputs\n\nAsk at most two questions, and only for what is missing:\n\n- **Audience.** For example a CIO, a CISO, a board or a procurement team.\n- **Decision.** For example choose a product, narrow a list, keep the current product or migrate.\n- **Scope.** A market, or two or three named products, and the place: a country, a region or \"world\".\n\n## Steps\n\nGather the facts with as few calls as possible. Deeper views use the user's Sovrnti credits.\n\n- **A whole market:** `sovrnti_market_get_snapshot`. Add `sovrnti_market_get_sovrntimax` if sovereignty is a hard requirement.\n- **Named products:** `sovrnti_product_compare`.\n- **A specific \"why\" question:** `sovrnti_qq_sherpa`. It answers only from what Sovrnti has scored and returns citations.\n\nUse the same place in every call. If results from earlier in the conversation already cover the scope, reuse them.\n\n## Output\n\nKeep it to about 400 words unless the user asks otherwise.\n\n1. **Title and date.**\n2. **The decision,** in one line.\n3. **Options,** one line each with its trade-off. Recommend one only if the results clearly support it, and say why.\n4. **Evidence.** A short table with the scores the tools returned.\n5. **Risks and what to verify.** Contract terms, configuration choices, and anything the results do not cover.\n6. **Sources.** \"Source: Sovrnti, <date>\", with the views used.\n\n## Rules\n\n- Use only facts the tools return. If the brief includes your own reading of the results, label it as your assessment, not as Sovrnti's.\n- Sovrnti's terms allow short extracts in internal documents such as board papers, with \"Source: Sovrnti\" and the date. Do not reproduce full result sets for publication.\n- Use plain language. Put numbers before adjectives.\n- If a tool returns an error, say so and state what the brief could not cover.\n- If a tool says the user has no credits left, pass its message on as it is. Do not suggest a workaround.\n"
}