YeonGyu-Kim b513d6e462 fix(api): sanitize tuning params for reasoning models (o1/o3/grok-3-mini) ...

Reasoning models reject temperature, top_p, frequency_penalty, and
presence_penalty with 400 errors. Instead of letting these flow through
and returning cryptic provider errors, strip them silently at the
request-builder boundary.

is_reasoning_model() classifies: o1*, o3*, o4*, grok-3-mini.
stop sequences are preserved (safe for all providers).

Tests added:
- reasoning_model_strips_tuning_params: o1-mini strips all 4 params, keeps stop
- grok_3_mini_is_reasoning_model: classification coverage for grok-3-mini, o1,
  o3-mini, and negative cases (gpt-4o, grok-3, claude)

85 api lib tests passing, 0 failing.

2026-04-08 07:32:47 +09:00

..

api

fix(api): sanitize tuning params for reasoning models (o1/o3/grok-3-mini)

2026-04-08 07:32:47 +09:00

commands

feat: b5-markdown-fence — batch 5 wave 2

2026-04-07 15:19:45 +09:00

compat-harness

wip: plugins progress

2026-04-01 07:09:06 +00:00

mock-anthropic-service

feat(harness+usage): add auto_compact and token_cost parity scenarios

2026-04-03 22:41:42 +09:00

plugins

Resolve claw-code main merge conflicts

2026-04-06 07:16:57 +00:00

runtime

fix(dead-code): remove stale constants + dead function; add workspace_sessions_dir tests

2026-04-08 04:04:54 +09:00

rusty-claude-cli

feat(api): add tuning params (temperature, top_p, penalties, stop) to MessageRequest

2026-04-08 07:07:33 +09:00

telemetry

feat: anthropic SDK header matching + request profile

2026-04-01 05:55:25 +00:00

tools

feat(api): add tuning params (temperature, top_p, penalties, stop) to MessageRequest

2026-04-08 07:07:33 +09:00