champollion 0.4.0 → 0.5.0

package/shared/licenses.json

@@ -1,7 +1,7 @@
 {
   "_generated": {
     "generatedBy": "mt-eval-arena/scripts/build-shared-licenses.mjs",
-    "generatedAt": "2026-08-18T20:28:43.461Z",
+    "generatedAt": "2026-10-05T19:07:50.966Z",
     "inputs": [
       "shared/license-register-floor.json (frozen champollion.db source_licenses, 2026-08-18 — human-verified terms incl. attribution/notes prose)",
       "docs/LICENSING.md (verified facts for supplemental entries)",
@@ -11,8 +11,8 @@
     "counts": {
       "fromFloor": 318,
       "supplemental": 17,
-      "fromAtlas": 0,
-      "total": 541
+      "fromAtlas": 8,
+      "total": 549
     },
     "note": "GENERATED FILE — do not edit by hand. Regenerate with: node mt-eval-arena/scripts/build-shared-licenses.mjs"
   },
@@ -6776,6 +6776,102 @@
       "non_commercial_only": 0,
       "dataset_url": null,
       "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "abvd-v0.1": {
+      "source": "abvd-v0.1",
+      "license_spdx": "CC-BY-4.0",
+      "license_url": null,
+      "attribution": "CLDF dataset derived from Greenhill et al.'s \"Austronesian Basic Vocabulary Database\" from 2020 — Simon J. Greenhill; Robert Forkel; Russell D. Gray; Bob Blust; Malcolm Ross; John Lynch; Bethwyn Evans; Erin Donnelly; Alex François; Emily Gasser; Mary Walworth; Benjamin Touati; Laurie Reid; Antoinette Schapper; Sander Adelaar; Isaac Stead; Tiago Tresoldi",
+      "allows_redistribution": 1,
+      "requires_attribution": 1,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "berrywestpapuan-v0.1": {
+      "source": "berrywestpapuan-v0.1",
+      "license_spdx": "CC-BY-4.0",
+      "license_url": null,
+      "attribution": "CLDF dataset derived from Berry and Berry's \"Survey of some West Papuan phylum languages\" from 1987 — Berry, Christine; Berry, Keith",
+      "allows_redistribution": 1,
+      "requires_attribution": 1,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "crossandean_morphology-v1.0": {
+      "source": "crossandean_morphology-v1.0",
+      "license_spdx": "CC-BY-4.0",
+      "license_url": null,
+      "attribution": "CLDF dataset on Quechua morphology derived from Blum et al.'s \"Una aproximación filolingüística a la clasificación interna del Quechua\" from 2021 — Frederic Blum; Carlos B. Ugarte; Adriano Ingunza; Zoe Poirier",
+      "allows_redistribution": 1,
+      "requires_attribution": 1,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "magram-v1.0": {
+      "source": "magram-v1.0",
+      "license_spdx": "CC-BY-4.0",
+      "license_url": null,
+      "attribution": "MAGRAM grammaticalization database — Walter Bisang; Andrej Malchukov; Marvin Martiny",
+      "allows_redistribution": 1,
+      "requires_attribution": 1,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "pachechibchan-v0.1": {
+      "source": "pachechibchan-v0.1",
+      "license_spdx": "CC-BY-4.0",
+      "license_url": null,
+      "attribution": "CLDF dataset derived from Pache's \"Pech and the basic internal classification of Chibchan\" from 2023 — Pache, Matthias; List, Johann-Mattis",
+      "allows_redistribution": 1,
+      "requires_attribution": 1,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "sabor-v1.0": {
+      "source": "sabor-v1.0",
+      "license_spdx": "CC-BY-4.0",
+      "license_url": null,
+      "attribution": "CLDF dataset accompanying Miller and List's \"Borrowing in South American Languages\" from 2023 — John Miller; Johann-Mattis List",
+      "allows_redistribution": 1,
+      "requires_attribution": 1,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "ud-treebanks-2.18": {
+      "source": "ud-treebanks-2.18",
+      "license_spdx": "LicenseRef-UD-Per-Treebank",
+      "license_url": null,
+      "attribution": "Universal Dependencies 2.18 treebank catalogue (Zeman et al., LINDAT/CLARIAH-CZ), release metadata retrieved 2026-08-18.",
+      "allows_redistribution": 0,
+      "requires_attribution": 0,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
+    },
+    "common-voice-cv-corpus-26.0-2026-06-12": {
+      "source": "common-voice-cv-corpus-26.0-2026-06-12",
+      "license_spdx": "MPL-2.0",
+      "license_url": null,
+      "attribution": "Mozilla Common Voice dataset statistics, cv-corpus-26.0-2026-06-12 (common-voice/cv-dataset), retrieved 2026-08-18.",
+      "allows_redistribution": 1,
+      "requires_attribution": 0,
+      "requires_sharealike": 0,
+      "non_commercial_only": 0,
+      "dataset_url": null,
+      "notes": "atlas cldf_sources — declared by the source SNAPSHOT at fetch time"
     }
   }
 }

package/shared/retired-model-aliases.json

@@ -0,0 +1,9 @@
+{
+  "_comment": "RETIRED — founder ruling 2026-10-05: every model is named by its exact provider slug; no alias resolves to a model. These short names are kept ONLY so the refusal can say which exact slug the old name used to stand for. Nothing may read this file to map a name to a model: an input that matches a key here is REFUSED (cli/lib/models.js requireExactModelId, arena/mt_eval_harness/config.py exact_model_refusal).",
+  "retired": {
+    "gemini-flash": "google/gemini-3.5-flash",
+    "gemini-pro": "google/gemini-3.1-pro-preview",
+    "claude-sonnet": "anthropic/claude-sonnet-4",
+    "gpt": "openai/gpt-5.5"
+  }
+}

package/shared/schemas/human-services.schema.json

@@ -2,7 +2,7 @@
   "$schema": "http://json-schema.org/draft-07/schema#",
   "$id": "https://champollion.dev/schemas/human-services.schema.json",
   "title": "Champollion Human Translation Services Registry (SSOT)",
-  "description": "Public, opt-in registry of human translation providers keyed by language pair — the in-git SSOT for both the CLI and the website (mirrors the model-aliases.json dual-runtime pattern). It is uploaded to public.translation_services (migration 034). PII (the provider `contact` address/endpoint) is NEVER carried here — it is supplied to the uploader out-of-band. See docs/HUMAN_SERVICES_HUB.md.",
+  "description": "Public, opt-in registry of human translation providers keyed by language pair — the in-git SSOT for both the CLI and the website (mirrors the retired-model-aliases.json dual-runtime pattern). It is uploaded to public.translation_services (migration 034). PII (the provider `contact` address/endpoint) is NEVER carried here — it is supplied to the uploader out-of-band. See docs/HUMAN_SERVICES_HUB.md.",
   "type": "array",
   "items": { "$ref": "#/$defs/service" },
   "$defs": {

package/shared/schemas/retired-model-aliases.schema.json

@@ -0,0 +1,32 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://champollion.dev/schemas/retired-model-aliases.schema.json",
+  "title": "Champollion retired model aliases (shared/retired-model-aliases.json)",
+  "description": "Short model names Champollion USED to accept, each with the exact provider slug it used to stand for. Founder ruling 2026-10-05: exact slugs only, no aliasing — every name in `retired` is REFUSED with a message naming its slug; it never resolves. The bundled mini-schema validator does not support additionalProperties-as-schema, so per-entry validation iterates `retired` against $defs.aliasName / $defs.modelId (see cli/test/shared-ssot-schemas.test.js).",
+  "type": "object",
+  "required": ["_comment", "retired"],
+  "properties": {
+    "_comment": {
+      "type": "string",
+      "description": "Human note: why these names are retired."
+    },
+    "retired": {
+      "type": "object",
+      "description": "Retired short name → the exact slug it used to stand for (used ONLY in the refusal message)."
+    }
+  },
+  "$defs": {
+    "aliasName": {
+      "type": "string",
+      "description": "Retired short name: lowercase alphanumerics and hyphens.",
+      "pattern": "^[a-z0-9][a-z0-9-]*$",
+      "minLength": 2
+    },
+    "modelId": {
+      "type": "string",
+      "description": "Exact provider-qualified model slug: <provider>/<model>, never a floating id (no leading '~', no '-latest').",
+      "pattern": "^[a-z0-9-]+/[A-Za-z0-9._-]+$",
+      "minLength": 3
+    }
+  }
+}

package/shared/model-aliases.json

@@ -1,7 +0,0 @@
-{
-  "_comment": "Convenience aliases. Full model list fetched from OpenRouter at runtime.",
-  "gemini-flash": "google/gemini-3.5-flash",
-  "gemini-pro": "google/gemini-3.1-pro-preview",
-  "claude-sonnet": "anthropic/claude-sonnet-4",
-  "gpt": "openai/gpt-5.5"
-}

package/shared/schemas/model-aliases.schema.json

@@ -1,27 +0,0 @@
-{
-  "$schema": "https://json-schema.org/draft/2020-12/schema",
-  "$id": "https://champollion.dev/schemas/model-aliases.schema.json",
-  "title": "Champollion model aliases (shared/model-aliases.json)",
-  "description": "Cross-runtime SSOT mapping short convenience aliases (e.g. 'gemini-flash') to full provider-qualified model ids (e.g. 'google/gemini-3.5-flash'). Keys other than the underscore-prefixed metadata fields are alias names; their values are model ids. The bundled mini-schema validator does not support additionalProperties-as-schema, so per-entry validation iterates the entries against $defs.aliasName / $defs.modelId (see cli/test/shared-ssot-schemas.test.js).",
-  "type": "object",
-  "properties": {
-    "_comment": {
-      "type": "string",
-      "description": "Human note; not an alias."
-    }
-  },
-  "$defs": {
-    "aliasName": {
-      "type": "string",
-      "description": "Short alias key: lowercase alphanumerics and hyphens.",
-      "pattern": "^[a-z0-9][a-z0-9-]*$",
-      "minLength": 2
-    },
-    "modelId": {
-      "type": "string",
-      "description": "Full OpenRouter-style model id: <provider>/<model>.",
-      "pattern": "^[a-z0-9-]+/[A-Za-z0-9._-]+$",
-      "minLength": 3
-    }
-  }
-}