Update to Slop Curator

{
  "name": "slop-quality-gate",
  "description": "Check an edited or newly written draft for factual drift, voice loss, over-editing, residual formulaic patterns, robotic rhythm, sayability failures, and output-contract failures. Use after a rewrite, a plain-spoken pass, or when the user wants a quality check.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 313
    }
  ],
  "skill_md_contents": "---\nname: slop-quality-gate\ndescription: Check an edited or newly written draft for factual drift, voice loss, over-editing, residual formulaic patterns, robotic rhythm, sayability failures, and output-contract failures. Use after a rewrite, a plain-spoken pass, or when the user wants a quality check.\n---\n\n# Slop Quality Gate\n\nEvaluate the finished draft independently. Do not reward the edit merely for being shorter or more polished.\n\n## Hard gates\n\nFail immediately if the edit:\n\n- invents or changes a fact, number, date, quote, source, name, or claim\n- asserts that AI wrote the text\n- changes the user's intended position\n- removes a meaningful constraint or qualification\n- flattens a distinctive voice into generic professional prose\n- turns requested spoken or dialect writing into formal article prose\n- fabricates slang, personal stories, uncertainty, or other \"human\" details\n- invents proof, scarcity, urgency, testimonials, rankings, capabilities, or unsupported precision\n- violates an explicitly active house-style hard rule\n\n## Scorecard\n\nScore internally out of 100:\n\n- factual and semantic fidelity: 25\n- voice retention: 20\n- specificity and concrete support: 15\n- natural rhythm: 15\n- directness: 10\n- structural variety without theatrics: 10\n- formatting restraint: 5\n\nWhen `commercial-copy-director` was used, evidence integrity and truthful offer framing are hard requirements regardless of the numeric score. When `strict-human-output` house mode was active, active hard bans must pass.\n\nWhen `plain-spoken-writing` was part of the route, treat sayability, natural thought order, dialect authenticity, and code-switching fidelity as part of voice retention and natural rhythm. Do not reward grammatical polish that makes the result less speakable.\n\nDefault pass threshold: 85, with no hard-gate failure.\n\nDo not show the numeric score unless the user asks for scoring.\n\n## Residual checks\n\nCheck for:\n\n- repeated binary reframes\n- repeated throat-clearing or faux-insight setups\n- unsupported authority\n- vague importance claims\n- synonym cycling\n- fake-profound ending\n- repeated question-answer scaffolds\n- stacked punchy fragments\n- excessive short-line stacking or micro-paragraphing\n- uniform paragraph geometry\n- formatting used only for emphasis\n- second-order anti-slop artifacts such as forced slang or over-compression\n\n## Feedback routing\n\nIf the draft fails, return only the failing regions to the responsible pass:\n\n- meaning, evidence, factual locks, or general existing-writer voice -> `voice-preserving-edit`\n- sayability, spoken thought order, Egyptian dialect, natural code-switching, or controlled roughness -> `plain-spoken-writing`\n- phrase, structure, rhythm, agency, or formatting -> `slop-pattern-repair`\n- Arabic dialect, Arabic syntax, code-switch terminology, or house Arabic vocabulary -> `arabic-style-curator`\n- active hard policy, evidence integrity, format compliance, or punctuation contract -> `strict-human-output`\n- headline, CTA, offer framing, proof, or commercial-decision failure -> `commercial-copy-director`\n- presentation/document hierarchy, fake visual content, RTL design, or asset-control failure -> `visual-content-anti-slop`\n\nRecheck the revised regions and then the full piece once. Avoid endless polishing loops.\n\n## Output behavior\n\nWhen invoked as part of another workflow, keep the gate internal and return the finished draft. When the user explicitly asks for a quality review, return pass/fail by category with concise evidence.\n"
}