Provider data pipeline: registry tiers, generated model catalogue, maintenance jobs, RFC-0033

.github/workflows/ci.yml

@@ -398,6 +398,10 @@ jobs:
       - uses: actions/checkout@v4
       - name: ProviderName generator drift (checked-in files must match generator output)
         run: python3 scripts/gen_provider_names.py --check
+      - name: Provider doc drift (docs/api/providers.md must match generator output)
+        run: python3 scripts/gen_providers_doc.py --check
+      - name: Model data drift (aimux-providers/data must match its manifest)
+        run: python3 scripts/gen_models.py --check
       - uses: dtolnay/rust-toolchain@stable
       - uses: Swatinem/rust-cache@v2
       - uses: actions/setup-node@v4

.github/workflows/registry-maintenance.yml

@@ -0,0 +1,71 @@
+# Provider data maintenance (RFC-0033 §4.1). Nothing here edits
+# provider_registry.json — the reports land on issue #170 for a human, and the
+# generated L2 data lands on a PR.
+name: Registry maintenance
+
+on:
+  schedule:
+    - cron: "17 5 * * 1" # Mondays — upstream registry diff
+    - cron: "17 6 * * 1" # Mondays — refresh the generated model data
+    - cron: "17 7 1 * *" # 1st of the month — reachability probe
+  workflow_dispatch:
+
+permissions:
+  contents: write
+  issues: write
+  pull-requests: write
+
+env:
+  GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+jobs:
+  sync-report:
+    name: Upstream registry diff
+    if: github.event_name == 'workflow_dispatch' || github.event.schedule == '17 5 * * 1'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - run: |
+          {
+            echo "### Upstream registry diff — $(date -u +%Y-%m-%d)"
+            echo
+            python3 scripts/sync_registry.py --report
+          } > sync-report.md
+      - run: gh issue comment 170 --body-file sync-report.md
+
+  refresh-models:
+    name: Refresh model data
+    if: github.event_name == 'workflow_dispatch' || github.event.schedule == '17 6 * * 1'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - run: python3 scripts/gen_models.py
+      - name: Commit and open/update the PR
+        run: |
+          if git diff --quiet aimux-providers/data; then
+            echo "model data unchanged"; exit 0
+          fi
+          git config user.name "github-actions[bot]"
+          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
+          git checkout -B bot/refresh-model-data
+          git add aimux-providers/data
+          git commit -m "chore(data): refresh the generated model catalogue"
+          git push --force origin bot/refresh-model-data
+          gh pr create --base master --head bot/refresh-model-data \
+            --title "chore(data): refresh the generated model catalogue" \
+            --body "Automated \`scripts/gen_models.py\` run. Review the manifest counts, then merge." \
+            || gh pr edit bot/refresh-model-data --body "Automated \`scripts/gen_models.py\` run, refreshed $(date -u +%Y-%m-%d)."
+
+  probe:
+    name: Reachability probe
+    if: github.event_name == 'workflow_dispatch' || github.event.schedule == '17 7 1 * *'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - run: |
+          {
+            echo "### Registry reachability probe — $(date -u +%Y-%m-%d)"
+            echo
+            python3 scripts/probe_registry.py --timeout 8 --concurrency 16
+          } > probe-report.md
+      - run: gh issue comment 170 --body-file probe-report.md

README.md

@@ -340,6 +340,7 @@ Tests run on cassette playback — no network and no keys. See
 | [docs/api/reference.md](docs/api/reference.md) | **API reference** — public types & functions lookup |
 | [docs/api/providers.md](docs/api/providers.md) | **Provider list** — all 325 providers with entry points (generated) |
 | [docs/api/](docs/api/) | **Per-language API guides** — Node.js, Python, Rust, Go, C/C++, Swift, Kotlin, Flutter |
+| [docs/contributing/adding-a-provider.md](docs/contributing/adding-a-provider.md) | **Adding a provider** — the four-layer model and a checklist per case |
 | [docs/error-model.md](docs/error-model.md) | **错误模型** — 跨语言错误形态与兼容性约定 |
 | [docs/PROJECT-OVERVIEW.md](docs/PROJECT-OVERVIEW.md) | Project overview, design decisions, benchmarks |
 | [docs/PERF-RESULTS.md](docs/PERF-RESULTS.md) | Performance benchmark results |
@@ -379,6 +380,7 @@ Tests run on cassette playback — no network and no keys. See
 | [0027](rfc/0027-model-catalogue-and-list-api.md) | Model catalogue & `list_models` |
 | [0028](rfc/0028-transcription-streaming.md) | Realtime transcription streaming (WS sessions) |
 | [0029](rfc/0029-web-console.md) | `aimux-web` browser console |
+| [0033](rfc/0033-code-convergence-plan.md) | Code convergence plan (providers / FFI / bindings / recording / errors / tests / docs) |
 
 ## Contributing
 

aimux-providers/Cargo.toml

@@ -10,7 +10,7 @@ documentation = "https://docs.rs/aimux-providers"
 # Cassette recordings for the integration tests are huge (88MB); they are not
 # needed by downstream consumers, so keep them out of the published crate
 # (crates.io limit is 10MB).
-exclude = ["tests/cassettes/**", "tests/cassettes"]
+exclude = ["tests/cassettes/**", "tests/cassettes", "data/**", "data"]
 
 [dependencies]
 aimux-core = { workspace = true }

aimux-providers/data/models.overrides.json

@@ -0,0 +1,3 @@
+{
+  "_doc": "Hand-maintained corrections, applied by scripts/gen_models.py after upstream normalisation. Shape: { \"<provider>\": { \"<model id>\": { <fields to override> } } }, or \"_drop\": true to remove a model. Example: { \"groq\": { \"llama-3.3-70b-versatile\": { \"max_output\": 32768 }, \"whisper-large-v3\": { \"_drop\": true } } }. Top-level keys starting with '_' are ignored (no provider is named that way)."
+}

aimux-providers/data/models/.manifest.json

@@ -0,0 +1,11 @@
+{
+  "schema_version": 1,
+  "generated_at": "2026-09-04T13:09:34Z",
+  "sources": {
+    "models.dev": "https://models.dev/api.json",
+    "anya2a": "https://raw.githubusercontent.com/ThinkInAIXYZ/PublicProviderConf/refs/heads/dev/dist/all.json"
+  },
+  "providers": 137,
+  "models": 4872,
+  "hash": "566a87de7cf1a047698ffad95701da90c9e44e9e478b1ccae5cb6a1a71f6055a"
+}