teknium1 749f657fbf fix(stt): coerce the new client keys, apply the 60s default to the QQ STT call, document both ...

Quoted YAML numerics ("60") reached the SDK as strings; read stt.openai.timeout /
max_retries through _config_number like every other numeric stt key. The QQ
adapter's OpenAI-compatible STT call carried the same fixed 30s and now honours
platforms.qqbot.extra.stt.timeout (default 60) so a cold-starting self-hosted
model is not cut off there either. Document the keys and the trade-off.

Co-authored-by: Paolo Antinori <pantinor@redhat.com>

2026-09-19 10:12:15 -07:00

..

developer-guide

fix: strip stale Codex reasoning on 401 token_expired before the credential pool

2026-09-19 09:32:00 -07:00

getting-started

fix(update): "Already up to date" re-syncs a venv installed from an older release

2026-09-19 02:02:14 -07:00

guides

fix(picker): Entra-only Azure Foundry stays listed in /model without an API key

2026-09-19 09:42:49 -07:00

integrations

docs: note that session-less one-shots send an ephemeral x-opencode-session key

2026-09-19 10:06:00 -07:00

reference

fix(gateway): a --global /model keeps the session override under a channel_overrides model; report a failed stale-override cleanup

2026-09-19 09:59:32 -07:00

user-guide

fix(stt): coerce the new client keys, apply the 60s default to the QQ STT call, document both

2026-09-19 10:12:15 -07:00

index.mdx

docs(website): link pages by relative Markdown path so they open on GitHub (#114428)

2026-09-18 14:27:04 -07:00

user-stories.mdx

…