You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Harness: provider form for extra headers, and before/after measurement for #262 #267
Part of #261. Follow-up to #262, which #266 merged except for these two parts.
Settings form for extra headers and query parameters
#266 lets OpenAI-format providers carry extraHeaders and queryParams (P-NET-11): the server stores them, sends them upstream, and rejects them for Anthropic-format providers. The desktop provider form does not offer them yet, so today they can only be set by editing providers.json. Add them to the provider form (desktop and H5 where providers are edited), with i18n in both languages, for example for Azure's api-version or OpenRouter's attribution headers.
Before and after measurement
#262 asked for a before/after comparison on the same task set: cache hit tokens, input tokens and turns per task. #266 measured only cache hits on multi-turn requests through sendToUpstream (gpt-4.1-mini about 99% from turn 2, glm-4.6 13312 of 13371 input tokens on turn 3, on OpenRouter). Run the task set on the commit before #266 and on main after it, with the same providers, and record the three numbers in docs/features/harness.
Activity