ef0252bb3f
fix: improve smoke test prompt for reliable tool calling
...
The previous prompt 'please list files in the current directory' was
ambiguous and didn't explicitly require tool usage. Models like
qwen/qwen3-coder and z-ai/glm-4.6 would sometimes respond with text
describing what they would do instead of actually calling the tool.
The new prompt explicitly instructs the model to immediately call the
shell tool without asking for confirmation, which should improve
reliability for models with weaker tool-calling capabilities.
2025-12-24 20:39:09 -08:00
5ca7eb2305
chore: Update gemini versions in test_providers.sh ( #6246 )
2025-12-23 11:12:19 +11:00
7134e89c4b
feat: improved UX for tool calls via execute_code ( #6205 )
2025-12-22 10:42:20 -05:00
d4814042e6
chore: cover code mode with end to end provider tests ( #6183 )
2025-12-19 12:02:06 +08:00
cfdf01567d
fix: support Gemini 3's thought signatures ( #5806 )
...
Signed-off-by: Salvatore Testa <sal@withpersona.com >
2025-11-20 16:28:27 +11:00
2bef034303
feat: trying grok for live test ( #5732 )
2025-11-17 09:37:43 +11:00
89f7384d57
add clippy warning for string_slice ( #5422 )
...
Co-authored-by: Douwe Osinga <douwe@squareup.com >
2025-11-04 17:46:25 -05:00
7511a533d6
we should run this on main and also test open models at least via ope… ( #5556 )
...
adds qwen3-code and GLM 4.6 to test_providers for open model coverage
2025-11-04 09:06:23 +11:00
38e7dc8f30
fix: remove qwen3-coder from provider/mcp smoke tests ( #5551 )
2025-11-03 14:33:49 -05:00
b94535b679
testing tetrate with sonnet ( #5428 )
2025-10-29 11:40:02 +11:00
6b6c50976c
Gemini again ( #5390 )
...
Co-authored-by: Douwe Osinga <douwe@squareup.com >
2025-10-27 16:41:00 -04:00
3c975bb358
live testing script ( #5263 )
...
Co-authored-by: Jack Amadeo <jackamadeo@squareup.com >
2025-10-21 16:39:58 +11:00
64b37339e0
Skip subagents for gemini ( #5257 )
...
Co-authored-by: Douwe Osinga <douwe@squareup.com >
2025-10-18 17:35:29 -04:00
Bradley Axen