oterm/tests/unit
Yiorgis Gozadinos 63005bb801
Filter chat parameters against the provider's supported settings
_build_model_settings now consults get_supported_setting_keys(provider)
instead of a hard-coded three-key allowlist, so seed and the rest of the
common pydantic-ai sampling fields round-trip while stale Ollama-native
keys (num_ctx, repeat_penalty, …) are dropped silently.
2026-04-26 22:43:47 +03:00
..
__init__.py Add unit-tier tests for agent, store, config, utils, providers 2026-04-24 20:06:56 +03:00
test_agent.py Filter chat parameters against the provider's supported settings 2026-04-26 22:43:47 +03:00
test_capabilities.py Surface Ollama show_model failures to oterm.log 2026-04-26 13:02:53 +03:00
test_config.py Clean up stale config fields, CSS classes, and command descriptions 2026-04-25 00:06:16 +03:00
test_provider_settings.py Add provider→ModelSettings registry 2026-04-26 22:41:06 +03:00
test_providers.py Unify env-var expansion on ${VAR} syntax across config 2026-04-26 20:28:36 +03:00
test_providers_ollama.py Route Ollama through OpenAIChatModel and drop parameter pre-fill 2026-04-26 22:38:33 +03:00
test_store.py Rewrite tools-tier tests around TestModel, stdio MCP, no VCR 2026-04-24 20:20:37 +03:00
test_store_upgrades.py Cover remaining branches across modals, widgets, store, and MCP setup 2026-04-25 23:43:27 +03:00
test_types.py Yield (thinking, text) tuples from stream_agent 2026-04-24 22:11:03 +03:00
test_utils.py Unify env-var expansion on ${VAR} syntax across config 2026-04-26 20:28:36 +03:00