{
  "briefs": [
    {
      "id": "brf_claude_task_brief",
      "name": "Briefing Claude on a task (prompt engineering)",
      "discipline": "prompt-engineering",
      "summary": "Anthropic's own prompting guidance converges on the same shape as the legal and marketing brief conventions above, applied to instructing Claude: state the task explicitly, name who the output is for, and define what \"done\" looks like before generation starts.",
      "structure_md": "- **Task** — name the task explicitly rather than leaving it implicit\n- **Audience** — identify who the output is for\n- **Format / length / purpose** — specific details up front, not left for Claude to infer\n- **Done criteria** — clear expectations for what a correct/complete output looks like, set before any generation begins",
      "source_doc": "https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-prompting-best-practices",
      "created_at": "2026-07-02 06:01:18",
      "cite_as": "https://subagentbriefs.com/api/briefs/brf_claude_task_brief"
    }
  ]
}