{
  "schema_version": "1.0",
  "receipt_id": "par_public_writing_2026_09_17",
  "need": "Keep the repository's most visible writing clear, credible, and interesting without making a model or a bespoke prose engine authoritative.",
  "artifact_role": "public_writing_quality_loop",
  "discovery": {
    "queries": [
      "offline prose linter Vale",
      "First Reader agent skill",
      "ZeroSlop AI writing patterns",
      "inclusive language linter alex",
      "custom readability script",
      "model-only rewrite"
    ],
    "stop_condition": "A mature deterministic tool covered repeatable wording checks, while small review-only techniques covered first-impression and AI-pattern judgment without entering runtime behavior."
  },
  "contenders": [
    {
      "name": "Vale 3.21.0 and vale-action 3.0.0",
      "disposition": "adopt",
      "evidence": [
        "MIT",
        "Vale tag commit ed7117697bd9e8e0f8cd6b6217125878321ef17b",
        "vale-action tag commit 518a9136acc6e6668ce7c00d367051e0941e87ff",
        "6108 and 249 GitHub stars respectively at 2026-09-17 review",
        "offline project-owned rules",
        "official pinned GitHub Action"
      ]
    },
    {
      "name": "First Reader in Shubhamsaboo/awesome-llm-apps at f163bb5a92111cee4610ac98e5dce4c6a2a09c26",
      "disposition": "trial",
      "evidence": [
        "Apache-2.0 parent repository",
        "fresh-reader quick skim separates first impression from editorial context",
        "no runtime or package dependency",
        "parent repository is a broad learning collection rather than a blanket dependency endorsement"
      ]
    },
    {
      "name": "ZeroSlop 2.12.1",
      "disposition": "reference",
      "evidence": [
        "MIT",
        "tag commit 98e34b5cddbafdb25bff385d046fe0cef4d05fad",
        "122 GitHub stars at 2026-09-17 review",
        "offline scorer is explicitly heuristic and not an authorship detector"
      ]
    },
    {
      "name": "reader-first-writing-skills",
      "disposition": "watch",
      "evidence": [
        "Apache-2.0",
        "useful fidelity and reader-journey concepts",
        "one GitHub star and no stable release at 2026-09-17 review"
      ]
    },
    {
      "name": "alex",
      "disposition": "reference",
      "evidence": [
        "MIT",
        "5101 GitHub stars at 2026-09-17 review",
        "valuable but narrower inclusive-language role",
        "last repository push observed in 2024"
      ]
    },
    {
      "name": "custom prose scorer or model-only rewrite gate",
      "disposition": "reject",
      "evidence": [
        "duplicates established linting mechanics",
        "opaque or nondeterministic pass criteria",
        "risk of changing factual meaning and repository voice"
      ]
    }
  ],
  "hard_gates": [
    "runtime and canonical state remain model-free",
    "automated prose checks are deterministic and offline",
    "rules target a small public entrypoint set rather than generated upstream descriptions",
    "human review remains responsible for meaning, evidence, and final tone",
    "AI-pattern scores never become authorship claims or release thresholds",
    "hosted rewriting services receive no unpublished repository prose"
  ],
  "unknowns": [
    "First-impression judgments vary by reader and cannot be reduced to a score.",
    "Project-owned Vale rules need evidence-based refinement as the public voice evolves.",
    "Young AI-tone tools may mature enough to justify a later bounded trial."
  ],
  "decision_effect": "Use Vale 3.21.0 with a pinned official action and narrow local rules for repeatable public-copy checks. Require a context-free first-reader skim for material entrypoint changes. Keep First Reader and ZeroSlop as optional review techniques only; neither can author facts, gate releases by score, or enter the runtime dependency graph.",
  "reconsider_when": [
    "Vale rules create repeated false positives or miss a measured public-writing failure",
    "A mature offline tool demonstrates better reader-journey checks with a clear removal boundary",
    "The public entrypoint set or documentation generator changes materially",
    "AI-assisted editing can no longer preserve an inspectable human approval boundary"
  ],
  "observed_at": "2026-09-17T16:00:00Z"
}