docs: de-tell contributing page; codify style rule in CLAUDE.md

CLAUDE.md

@@ -95,3 +95,22 @@ actual registered task names.
   rewrites GitHub-only alert syntax (`> [!NOTE]` etc.) in README.md into bold
   blockquotes (`> **Note:**`) that PyPI renders; keep using alert syntax in the
   README itself. Config lives at the bottom of pyproject.toml.
+
+## Writing style (public-facing text)
+
+READMEs, docs pages, repo/collection descriptions, and HF model cards must
+avoid AI-writing tells. This repo hosts the canonical rule: the full version
+with the gating checklist lives in
+[docs/model-cards.md, "Writing style"](docs/model-cards.md);
+short version:
+
+- No em dashes in prose. Use periods, colons, commas, or parentheses (`—` is
+  fine as an empty table cell and inside code blocks).
+- Bold only for definition-list lead-ins (`**term:**`) and at most one critical
+  imperative per safety bullet. Never mid-sentence for emphasis.
+- No decorative emoji (functional ✅/⚠️ marks and 🤗 for Hugging Face are fine),
+  no slogans or chiasmus, no "not just X, but Y".
+- Headers use colons, never em dashes or italics.
+
+Style-only edits must never touch YAML frontmatter, code blocks, numbers,
+links, or safety qualifiers.

docs/contributing.md

@@ -5,7 +5,7 @@ repository that:
 
 1. **depends on [Inspect Robots](https://github.com/robocurve/inspect-robots)**,
 2. **defines one or more Inspect Robots `Task`s**, and
-3. **registers them via entry points** — `[project.entry-points."inspect_robots.tasks"]`
+3. **registers them via entry points**: `[project.entry-points."inspect_robots.tasks"]`
    (and, if it ships a simulator/embodiment or policy,
    `inspect_robots.embodiments` / `inspect_robots.policies`).
 
@@ -33,10 +33,10 @@ Benchmark(
 )
 ```
 
-The homepage card grid is **generated from this catalog at build time**, so adding
+The homepage card grid is generated from this catalog at build time, so adding
 an entry updates the site automatically. A test validates every entry (unique
 name, well-formed `https://github.com/...` repo URL, ≥1 task key), and CI requires
-100% coverage — keep `task_keys` in sync with the benchmark's actual registered
+100% coverage. Keep `task_keys` in sync with the benchmark's actual registered
 task names.
 
 See the [API reference](api.md) for the `Benchmark` fields.

docs/model-cards.md

@@ -31,7 +31,7 @@ Cards are read as credibility signals: prose patterns that readers flag as AI-ge
 undermine the numbers next to them. Avoid the tells.
 
 - No em dashes in prose; use periods, colons, commas, or parentheses. `—` is allowed only
-  as an empty table cell.
+  as an empty table cell or inside a code block.
 - Bold only for definition-list lead-ins (`**term:**`) and at most one critical imperative
   per safety bullet; never mid-sentence for emphasis.
 - No decorative emoji (🤗 for Hugging Face is fine), no slogans or chiasmus, no

mkdocs.yml

@@ -1,6 +1,6 @@
 site_name: WorldEvals
 site_description: >-
-  The Inspect Evals for robotics — a curated catalog of physical-AI benchmarks
+  The Inspect Evals for robotics: a curated catalog of physical-AI benchmarks
   built on Inspect Robots.
 site_url: https://worldevals.org/
 repo_url: https://github.com/robocurve/worldevals

scripts/gen_catalog.py

@@ -1,6 +1,6 @@
 """Generate the WorldEvals homepage from the catalog at build time.
 
-Run automatically by the ``mkdocs-gen-files`` plugin during ``mkdocs build`` — it
+Run automatically by the ``mkdocs-gen-files`` plugin during ``mkdocs build``: it
 imports :data:`worldevals.catalog.CATALOG` and renders it as a Material grid-card
 listing, so the site is always in sync with ``catalog.py`` (the single source of
 truth), mirroring how Inspect Evals generates its listing from per-eval metadata.
@@ -55,20 +55,20 @@ def render_home() -> str:
 # WorldEvals
 
 <p style="font-size: 1.25rem; font-weight: 500; margin-bottom: 0.25rem;">
-The <strong>Inspect Evals</strong> for robotics.
+The Inspect Evals for robotics.
 </p>
 
 A curated catalog of physical-AI / VLA benchmarks built on
-[Inspect Robots](https://github.com/robocurve/inspect-robots). Each benchmark lives in **its
-own repository** (so it owns its release cadence, dependencies, and hardware
+[Inspect Robots](https://github.com/robocurve/inspect-robots). Each benchmark lives in its
+own repository (so it owns its release cadence, dependencies, and hardware
 notes); WorldEvals is the index that ties them together.
 
 [:octicons-mark-github-16: GitHub](https://github.com/robocurve/worldevals){{ .md-button }}
 [Add a benchmark](contributing.md){{ .md-button }}
 
 !!! tip "Two views of the collection"
-    `inspect-robots list` tells you what Inspect Robots tasks are **installed**.
-    `worldevals list` tells you what benchmarks **exist** and how to get them.
+    `inspect-robots list` tells you what Inspect Robots tasks are installed.
+    `worldevals list` tells you what benchmarks exist and how to get them.
 
 ## Benchmarks