docs: clarify role model overrides

This commit is contained in:
Eyejoker
2026-05-29 04:31:35 +09:00
committed by GitHub
parent 96b800e812
commit e17e5259d5
2 changed files with 8 additions and 5 deletions

View File

@@ -13,6 +13,7 @@ DISCORD_ARBITER_BOT_TOKEN= # Arbiter bot
# --- Agent type configuration --- # --- Agent type configuration ---
OWNER_AGENT_TYPE=codex # codex | claude-code OWNER_AGENT_TYPE=codex # codex | claude-code
REVIEWER_AGENT_TYPE=claude-code # claude-code | codex REVIEWER_AGENT_TYPE=claude-code # claude-code | codex
# ARBITER_AGENT_TYPE=codex # optional: codex | claude-code
# --- API keys --- # --- API keys ---
CLAUDE_CODE_OAUTH_TOKENS= # Canonical comma-separated Claude Code OAuth tokens (use one value for single-account too) CLAUDE_CODE_OAUTH_TOKENS= # Canonical comma-separated Claude Code OAuth tokens (use one value for single-account too)
@@ -85,8 +86,8 @@ STATUS_CHANNEL_ID= # Discord channel ID for live status updat
# REVIEWER_EFFORT=high # Reviewer effort override # REVIEWER_EFFORT=high # Reviewer effort override
# REVIEWER_FALLBACK_ENABLED=true # Fall back to codex on Claude failure (default: true) # REVIEWER_FALLBACK_ENABLED=true # Fall back to codex on Claude failure (default: true)
# #
# ARBITER_MODEL=claude-opus-4-8 # Arbiter model override # ARBITER_MODEL=gpt-5.5 # Arbiter model override when ARBITER_AGENT_TYPE=codex
# ARBITER_EFFORT=high # Arbiter effort override # ARBITER_EFFORT=xhigh # Arbiter effort override
# ARBITER_FALLBACK_ENABLED=true # Fall back to codex on Claude failure (default: true) # ARBITER_FALLBACK_ENABLED=true # Fall back to codex on Claude failure (default: true)
# --- Mixture of Agents (MoA) --- # --- Mixture of Agents (MoA) ---

View File

@@ -27,7 +27,7 @@ DISCORD_ARBITER_BOT_TOKEN=
# 역할별 provider # 역할별 provider
OWNER_AGENT_TYPE=codex OWNER_AGENT_TYPE=codex
REVIEWER_AGENT_TYPE=claude-code REVIEWER_AGENT_TYPE=claude-code
# ARBITER_AGENT_TYPE=claude-code ARBITER_AGENT_TYPE=codex
# provider 기본 모델 # provider 기본 모델
# 공식 Claude API 모델 ID입니다. # 공식 Claude API 모델 ID입니다.
@@ -46,14 +46,16 @@ REVIEWER_MODEL=claude-opus-4-8
REVIEWER_EFFORT=high REVIEWER_EFFORT=high
REVIEWER_FALLBACK_ENABLED=true REVIEWER_FALLBACK_ENABLED=true
ARBITER_MODEL=claude-opus-4-8 ARBITER_MODEL=gpt-5.5
ARBITER_EFFORT=high ARBITER_EFFORT=xhigh
ARBITER_FALLBACK_ENABLED=true ARBITER_FALLBACK_ENABLED=true
``` ```
설명: 설명:
- reviewer / arbiter provider 선택은 **전역 설정** - reviewer / arbiter provider 선택은 **전역 설정**
- 역할별 `*_MODEL`은 해당 역할의 provider에 맞춰 주입됩니다. `codex` 역할에는 Codex 모델을, `claude-code` 역할에는 Claude 모델을 넣어야 합니다.
- 예를 들어 `ARBITER_AGENT_TYPE=claude-code`로 바꾸면 그때 `ARBITER_MODEL=claude-opus-4-8`을 사용할 수 있습니다.
- 특정 방에서만 reviewer를 Codex로 고르는 기능은 없음 - 특정 방에서만 reviewer를 Codex로 고르는 기능은 없음
- room-level `agentConfig``claudeModel`, `codexModel`은 역할별이 아니라 provider별 override - room-level `agentConfig``claudeModel`, `codexModel`은 역할별이 아니라 provider별 override
- `ARBITER_AGENT_TYPE`은 옵션이며, 설정하지 않으면 arbiter는 비활성 상태입니다 - `ARBITER_AGENT_TYPE`은 옵션이며, 설정하지 않으면 arbiter는 비활성 상태입니다