From cca3b6c279b103f584a00977968155cf16d985eb Mon Sep 17 00:00:00 2001 From: nesquena-hermes Date: Sat, 4 Jul 2026 07:06:05 +0000 Subject: [PATCH] release #5511: exclude Copilot from providers: routing scan + regression tests + CHANGELOG MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Gate-cert (nesquena-hermes) + Codex found 1 CORE: akay64's new providers: scan in resolve_model_provider() iterated ALL providers..models INCLUDING providers.copilot.models — but that key is a documented per-model SETTINGS map (reasoning_effort/limits), NOT a routing allowlist (config.py exception). So a Copilot per-model settings entry hijacked that model's routing to Copilot (Codex reproduced gpt-5 -> copilot). Fix (Codex-exact): skip copilot in the scan. Added the regression tests the PR shipped 0 of: user-defined allowlist routes; active-openai + providers.copilot.models (dict + list shapes) stays openai (both verified to fail without the exclusion). Co-authored-by: akay64 --- CHANGELOG.md | 1 + api/config.py | 7 ++++ tests/test_model_resolver.py | 65 ++++++++++++++++++++++++++++++++++++ 3 files changed, 73 insertions(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index fa800554..ee1f6913 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -14,6 +14,7 @@ ### Fixed +- **Models from a user-defined provider in `config.yaml` now route correctly.** If you defined a custom provider under the `providers:` block with an explicit `models:` allowlist, a bare model id from that list wasn't matched by request-time routing — it silently fell back to the default provider and often 404'd. Resolution now scans your `providers:` allowlists (exact match, same as `custom_providers:`), so those models reach the right endpoint. Copilot is deliberately excluded from this scan because `providers.copilot.models` is a per-model settings map (reasoning effort, limits), not a routing allowlist. Thanks @akay64. (#5511) - **The transcript no longer scrolls up when the composer grows.** When you were pinned to the bottom of a conversation and the composer got taller — multi-line typing, `Shift+Enter`, or a multi-line / dictation paste — the transcript viewport shrank by the same amount and left you stranded a row or two above the bottom (it read as the chat "randomly scrolling up" during normal use). The transcript now re-pins to the bottom whenever the composer grows, but only when you were genuinely still pinned there — if you'd scrolled up to read history, nothing yanks you back down. (#5516, #5514, #5515) - **A deleted conversation no longer comes back as a read-only "ghost" after a restart.** Deleting a WebUI session removed its file but left a recoverable row in the agent's state database, so after a server restart and page reload the deleted conversation could reappear as a read-only "Agent" session still showing its old transcript. Deleted WebUI sessions are now recorded in a durable tombstone that the sidebar projection, startup backup-recovery, audit, and session-claim paths all honor, so a deleted session stays gone (the backend keeps returning the 404 the client relies on to clear the stale route). Genuine crash recovery is unaffected — a session whose sidecar is lost without a delete is still restored — imported external-channel (messaging) sessions are exempt and keep their transcript, and a deleted id is automatically un-tombstoned if the same session is later re-created or re-imported. Thanks @rodboev. (#5504, #5498) - **A drag that ends over the sidebar no longer switches conversations by accident.** Releasing the mouse (a stray `pointerup`) over a session row after a drag that began elsewhere in the main content area could switch you into that conversation unintentionally. Session switching now ignores a `pointerup` that isn't a genuine click on the row. Thanks @webtecnica. (#5487, #5462) diff --git a/api/config.py b/api/config.py index 712dff59..346e6525 100644 --- a/api/config.py +++ b/api/config.py @@ -2572,6 +2572,13 @@ def resolve_model_provider(model_id: str) -> tuple: for slug, pdef in providers_cfg.items(): if not isinstance(pdef, dict): continue + # Copilot is the documented exception: `providers.copilot.models` is + # a per-model SETTINGS map (reasoning_effort, limits, etc.), NOT a + # routable allowlist (see the exception at the catalog-build site). + # Scanning it here would let a Copilot per-model settings entry + # hijack that model's routing away from its real provider (#5511). + if _canonicalise_provider_id(slug) == "copilot": + continue p_models = pdef.get('models') if isinstance(p_models, list): for m in p_models: diff --git a/tests/test_model_resolver.py b/tests/test_model_resolver.py index df80b94c..740ef176 100644 --- a/tests/test_model_resolver.py +++ b/tests/test_model_resolver.py @@ -748,3 +748,68 @@ def test_bare_custom_provider_no_base_url_with_known_prefix_keeps_custom_and_ful ) assert model == 'google/gemma-2-9b' assert base_url is None + + +# ── providers: (config.yaml user-defined provider) scan (#5511) ───────────── + +def _resolve_with_providers(model_id, providers_cfg, *, provider=None, default=None): + """Helper: temporarily set config.cfg['providers'] + model, call resolve, restore.""" + old_cfg = dict(config.cfg) + model_cfg = {} + if provider: + model_cfg['provider'] = provider + if default: + model_cfg['default'] = default + config.cfg['model'] = model_cfg + config.cfg['providers'] = providers_cfg + try: + return config.resolve_model_provider(model_id) + finally: + config.cfg.clear() + config.cfg.update(old_cfg) + + +def test_providers_scan_routes_user_defined_allowlist_5511(): + """A user-defined providers..models allowlist routes a bare model id + to that provider (the feature #5511 adds).""" + model, provider, base_url = _resolve_with_providers( + 'my-model-1', + {'myprov': {'base_url': 'https://my.example/v1', 'models': ['my-model-1', 'my-model-2']}}, + provider='openai', + default='gpt-5', + ) + assert provider == 'myprov', f"user-defined provider allowlist must route; got {provider!r}" + assert model == 'my-model-1' + assert base_url == 'https://my.example/v1' + + +def test_providers_scan_skips_copilot_settings_map_5511(): + """providers.copilot.models is a per-model SETTINGS map, NOT a routable + allowlist — a Copilot per-model settings entry must NOT hijack routing away + from the model's real provider (#5511 gate-cert CORE finding).""" + model, provider, base_url = _resolve_with_providers( + 'gpt-5', + {'copilot': {'models': {'gpt-5': {'reasoning_effort': 'high'}}}}, + provider='openai', + default='gpt-5', + ) + assert provider == 'openai', ( + "Copilot settings-map entry must NOT hijack routing; " + f"gpt-5 must stay on openai, got {provider!r}" + ) + assert model == 'gpt-5' + + +def test_providers_scan_copilot_list_shape_also_skipped_5511(): + """Defense in depth: even if providers.copilot.models is a list shape, the + Copilot exclusion still prevents a routing hijack.""" + model, provider, base_url = _resolve_with_providers( + 'gpt-5', + {'copilot': {'models': ['gpt-5', 'gpt-5-mini']}}, + provider='openai', + default='gpt-5', + ) + assert provider == 'openai', ( + f"Copilot (list shape) must not hijack routing; got {provider!r}" + ) + assert model == 'gpt-5'