mirror of
https://github.com/Heretek-AI/heretek-openclaw-core.git
synced 2026-07-01 14:17:57 -04:00
John Doe
888 lines
29 KiB
YAML
888 lines
29 KiB
YAML
# ==============================================================================
|
|
# Heretek OpenClaw — LiteLLM Configuration v2.1
|
|
# ==============================================================================
|
|
# Configuration: AMD GPU + MiniMax Primary + z.ai Failover
|
|
# Passthrough Endpoints: Per-agent virtual models
|
|
#
|
|
# Each agent has a virtual model endpoint (agent/steward, agent/alpha, etc.)
|
|
# Users can reassign models via LiteLLM WebUI without changing openclaw.json
|
|
#
|
|
# DEFAULT MODEL: MiniMax MiniMax-M2.7 (via LiteLLM Gateway)
|
|
# FAILOVER: z.ai GLM-5.1 (Coding API)
|
|
#
|
|
# This file is mounted to the LiteLLM container at /app/config.yaml
|
|
# Version: 2.1.0
|
|
# Last Updated: 2026-03-31
|
|
#
|
|
# PROVIDER TEMPLATES: Additional provider configurations are available in:
|
|
# - config/providers/openai.yaml
|
|
# - config/providers/anthropic.yaml
|
|
# - config/providers/google.yaml
|
|
# - config/providers/ollama.yaml
|
|
# - config/providers/azure-openai.yaml
|
|
# - config/providers/xai.yaml
|
|
# - config/providers/combined-example.yaml
|
|
# ==============================================================================
|
|
|
|
model_list:
|
|
# ==========================================================================
|
|
# PRIMARY: MiniMax MiniMax-M2.7 (All Agents Default)
|
|
# ==========================================================================
|
|
- model_name: minimax/MiniMax-M2.7
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "MiniMax MiniMax-M2.7 - Primary model for all agents"
|
|
max_tokens: 128000
|
|
input_cost_per_token: 0.0000001
|
|
output_cost_per_token: 0.0000004
|
|
|
|
- model_name: minimax/MiniMax-M2.5
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.5
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "MiniMax MiniMax-M2.5 - Legacy fallback"
|
|
max_tokens: 128000
|
|
|
|
# ==========================================================================
|
|
# FAILOVER: z.ai GLM-5.1 (Coding API)
|
|
# ==========================================================================
|
|
- model_name: zai/glm-5-1
|
|
litellm_params:
|
|
model: openai/glm-5-1
|
|
api_key: os.environ/ZAI_API_KEY
|
|
api_base: os.environ/ZAI_API_BASE
|
|
model_info:
|
|
description: "z.ai GLM-5.1 - Coding API failover"
|
|
max_tokens: 128000
|
|
|
|
- model_name: zai/glm-5
|
|
litellm_params:
|
|
model: openai/glm-5
|
|
api_key: os.environ/ZAI_API_KEY
|
|
api_base: os.environ/ZAI_API_BASE
|
|
model_info:
|
|
description: "z.ai GLM-5 - Legacy failover"
|
|
max_tokens: 128000
|
|
|
|
- model_name: zai/glm-4
|
|
litellm_params:
|
|
model: openai/glm-4
|
|
api_key: os.environ/ZAI_API_KEY
|
|
api_base: os.environ/ZAI_API_BASE
|
|
model_info:
|
|
description: "z.ai GLM-4 - Alternative failover"
|
|
max_tokens: 128000
|
|
|
|
# ==========================================================================
|
|
# AGENT PASSTHROUGH ENDPOINTS (Virtual Models)
|
|
# ==========================================================================
|
|
# Each agent has a virtual model that defaults to MiniMax MiniMax-M2.5
|
|
# Users can reassign these via LiteLLM WebUI without changing openclaw.json
|
|
# ==========================================================================
|
|
|
|
# Steward (Orchestrator) - Manages collective workflow
|
|
- model_name: agent/steward
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Steward Agent - Orchestrator role"
|
|
agent_role: orchestrator
|
|
agent_id: steward
|
|
|
|
# Alpha (Triad) - Primary deliberator
|
|
- model_name: agent/alpha
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Alpha Agent - Triad role"
|
|
agent_role: triad
|
|
agent_id: alpha
|
|
|
|
# Beta (Triad) - Secondary deliberator
|
|
- model_name: agent/beta
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Beta Agent - Triad role"
|
|
agent_role: triad
|
|
agent_id: beta
|
|
|
|
# Charlie (Triad) - Final deliberator
|
|
- model_name: agent/charlie
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Charlie Agent - Triad role"
|
|
agent_role: triad
|
|
agent_id: charlie
|
|
|
|
# Examiner (Interrogator) - Challenges decisions
|
|
- model_name: agent/examiner
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Examiner Agent - Interrogator role"
|
|
agent_role: interrogator
|
|
agent_id: examiner
|
|
|
|
# Explorer (Scout) - Scans for opportunities
|
|
- model_name: agent/explorer
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Explorer Agent - Scout role"
|
|
agent_role: scout
|
|
agent_id: explorer
|
|
|
|
# Sentinel (Guardian) - Security and monitoring
|
|
- model_name: agent/sentinel
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Sentinel Agent - Guardian role"
|
|
agent_role: guardian
|
|
agent_id: sentinel
|
|
|
|
# Coder (Artisan) - Code generation and review
|
|
- model_name: agent/coder
|
|
litellm_params:
|
|
model: zai/glm-5-1
|
|
api_key: os.environ/ZAI_API_KEY
|
|
api_base: os.environ/ZAI_API_BASE
|
|
model_info:
|
|
description: "Coder Agent - Artisan role (GLM-4 via z.ai passthrough)"
|
|
agent_role: artisan
|
|
agent_id: coder
|
|
|
|
# Dreamer (Visionary) - Creative speculation, future planning
|
|
- model_name: agent/dreamer
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Dreamer Agent - Visionary role"
|
|
agent_role: visionary
|
|
agent_id: dreamer
|
|
|
|
# Empath (Diplomat) - Emotional intelligence, user rapport
|
|
- model_name: agent/empath
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Empath Agent - Diplomat role"
|
|
agent_role: diplomat
|
|
agent_id: empath
|
|
|
|
# Historian (Archivist) - Memory keeper, context preservation
|
|
- model_name: agent/historian
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Historian Agent - Archivist role"
|
|
agent_role: archivist
|
|
agent_id: historian
|
|
|
|
# Arbiter (Adjudicator) - Decision maker, conflict resolver
|
|
- model_name: agent/arbiter
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Arbiter Agent - Adjudicator role"
|
|
agent_role: adjudicator
|
|
agent_id: arbiter
|
|
|
|
# Catalyst (Accelerator) - Process optimization, workflow acceleration
|
|
- model_name: agent/catalyst
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Catalyst Agent - Accelerator role"
|
|
agent_role: accelerator
|
|
agent_id: catalyst
|
|
|
|
# Chronos (Timekeeper) - Time management, scheduling
|
|
- model_name: agent/chronos
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Chronos Agent - Timekeeper role"
|
|
agent_role: timekeeper
|
|
agent_id: chronos
|
|
|
|
# Coordinator (Integrator) - Cross-agent coordination
|
|
- model_name: agent/coordinator
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Coordinator Agent - Integrator role"
|
|
agent_role: integrator
|
|
agent_id: coordinator
|
|
|
|
# Echo (Communicator) - External communications, broadcasts
|
|
- model_name: agent/echo
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Echo Agent - Communicator role"
|
|
agent_role: communicator
|
|
agent_id: echo
|
|
|
|
# Habit Forge (Optimizer) - Habit formation, pattern tracking
|
|
- model_name: agent/habit-forge
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Habit Forge Agent - Optimizer role"
|
|
agent_role: optimizer
|
|
agent_id: habit-forge
|
|
|
|
# Metis (Strategist) - Strategic planning, tactical analysis
|
|
- model_name: agent/metis
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Metis Agent - Strategist role"
|
|
agent_role: strategist
|
|
agent_id: metis
|
|
|
|
# Nexus (Connector) - System integration, API management
|
|
- model_name: agent/nexus
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Nexus Agent - Connector role"
|
|
agent_role: connector
|
|
agent_id: nexus
|
|
|
|
# Perceiver (Sensor) - Data interpretation, pattern extraction
|
|
- model_name: agent/perceiver
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Perceiver Agent - Sensor role"
|
|
agent_role: sensor
|
|
agent_id: perceiver
|
|
|
|
# Prism (Analyzer) - Multi-perspective analysis
|
|
- model_name: agent/prism
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Prism Agent - Analyzer role"
|
|
agent_role: analyzer
|
|
agent_id: prism
|
|
|
|
# Sentinel Prime (Commander) - Lead security agent, threat coordination
|
|
- model_name: agent/sentinel-prime
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
model_info:
|
|
description: "Sentinel Prime Agent - Commander role"
|
|
agent_role: commander
|
|
agent_id: sentinel-prime
|
|
|
|
# ==========================================================================
|
|
# OLLAMA EMBEDDING (Local - AMD GPU)
|
|
# ==========================================================================
|
|
- model_name: ollama/nomic-embed-text
|
|
litellm_params:
|
|
model: ollama/nomic-embed-text-v2-moe
|
|
api_base: http://ollama:11434
|
|
model_info:
|
|
description: "Nomic Embed Text v2 MoE - 768 dimensions"
|
|
mode: embedding
|
|
embedding_dim: 768
|
|
|
|
# ==========================================================================
|
|
# MODEL ALIASES (Convenience)
|
|
# ==========================================================================
|
|
- model_name: default
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
|
|
- model_name: fast
|
|
litellm_params:
|
|
model: minimax/MiniMax-M2.7
|
|
api_key: os.environ/MINIMAX_API_KEY
|
|
api_base: os.environ/MINIMAX_API_BASE
|
|
|
|
- model_name: failover
|
|
litellm_params:
|
|
model: zai/glm-5-1
|
|
api_key: os.environ/ZAI_API_KEY
|
|
api_base: os.environ/ZAI_API_BASE
|
|
|
|
# ==============================================================================
|
|
# LiteLLM Settings
|
|
# ==============================================================================
|
|
litellm_settings:
|
|
# Default model for requests without model specification
|
|
default_model: minimax/MiniMax-M2.7
|
|
|
|
# Drop unsupported parameters
|
|
drop_params: true
|
|
|
|
# Logging verbosity
|
|
set_verbose: false
|
|
|
|
# Request timeout (seconds)
|
|
request_timeout: 300
|
|
|
|
# Retry configuration
|
|
num_retries: 3
|
|
retry_after: 2
|
|
|
|
# Caching
|
|
cache: true
|
|
cache_params:
|
|
ttl: 3600
|
|
redis_url: os.environ/REDIS_URL
|
|
|
|
# Callbacks for logging
|
|
callbacks: []
|
|
success_callback: ["prometheus", "langfuse", "log_cost"]
|
|
failure_callback: []
|
|
|
|
# ==============================================================================
|
|
# General Settings
|
|
# ==============================================================================
|
|
general_settings:
|
|
# Database connection
|
|
database_url: os.environ/DATABASE_URL
|
|
|
|
# Redis connection
|
|
redis_url: os.environ/REDIS_URL
|
|
|
|
# Master key for API authentication
|
|
master_key: os.environ/LITELLM_MASTER_KEY
|
|
|
|
# Salt for encryption
|
|
salt_key: os.environ/LITELLM_SALT_KEY
|
|
|
|
# Environment identifier
|
|
environment: heretek-openclaw
|
|
|
|
# Logging level
|
|
log_level: INFO
|
|
|
|
# UI settings
|
|
ui: true
|
|
ui_access_mode: admin
|
|
|
|
# Proxy settings
|
|
proxy_batch_write_at: 60
|
|
|
|
# ==============================================================================
|
|
# A2A (Agent-to-Agent) Protocol Settings
|
|
# ==============================================================================
|
|
a2a_settings:
|
|
# Enable A2A protocol
|
|
enabled: true
|
|
|
|
# Agent timeout (seconds)
|
|
agent_timeout: 300
|
|
|
|
# Session persistence backend
|
|
session_persistence: redis
|
|
|
|
# Session TTL (seconds)
|
|
session_ttl: 86400
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Standard A2A Endpoints
|
|
# -------------------------------------------------------------------------
|
|
agent_endpoint_format: "/v1/agents/{agent_name}"
|
|
send_message_endpoint: "/v1/agents/{agent_name}/send"
|
|
receive_message_endpoint: "/v1/agents/{agent_name}/receive"
|
|
health_check_endpoint: "/health"
|
|
health_check_timeout: 10
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Task Handoff (NEW)
|
|
# -------------------------------------------------------------------------
|
|
task_handoff:
|
|
enabled: true
|
|
context_preservation: full
|
|
timeout_seconds: 60
|
|
task_handoff_endpoint: "/v1/agents/{agent_name}/tasks"
|
|
task_status_endpoint: "/v1/agents/{agent_name}/tasks/{task_id}/status"
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Streaming Support (NEW)
|
|
# -------------------------------------------------------------------------
|
|
streaming:
|
|
enabled: true
|
|
chunk_size: 1024
|
|
streaming_endpoint: "/v1/agents/{agent_name}/stream"
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Agent Discovery (NEW)
|
|
# -------------------------------------------------------------------------
|
|
agent_discovery:
|
|
enabled: true
|
|
heartbeat_interval: 30
|
|
stale_agent_timeout: 120
|
|
agent_cards_endpoint: "/v1/agents/.well-known"
|
|
discover_endpoint: "/v1/agents/discover"
|
|
|
|
# ==============================================================================
|
|
# Router Settings
|
|
# ==============================================================================
|
|
router_settings:
|
|
# Routing strategy
|
|
routing_strategy: simple-shuffle
|
|
|
|
# Rate limiting
|
|
enable_rate_limiting: true
|
|
default_limit_requests: 60
|
|
default_limit_period: 60
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Model Health Checks
|
|
# -------------------------------------------------------------------------
|
|
model_health_check:
|
|
enabled: true
|
|
check_interval: 30
|
|
timeout: 10
|
|
unhealthy_threshold: 2
|
|
recovery_wait: 30
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Priority-based Fallback
|
|
# -------------------------------------------------------------------------
|
|
priority_fallback:
|
|
enabled: true
|
|
fallback_order:
|
|
- minimax/MiniMax-M2.7
|
|
- minimax/MiniMax-M2.5
|
|
- zai/glm-5-1
|
|
- zai/glm-5
|
|
|
|
# ==============================================================================
|
|
# Fallback Models (Priority-Ordered)
|
|
# ==============================================================================
|
|
fallback_models:
|
|
# Priority 1: Primary MiniMax
|
|
- minimax/MiniMax-M2.7
|
|
- minimax/MiniMax-M2.5
|
|
# Priority 2: z.ai Coding API
|
|
- zai/glm-5-1
|
|
- zai/glm-5
|
|
|
|
# ==============================================================================
|
|
# Per-Agent Model Mappings
|
|
# ==============================================================================
|
|
# This section defines per-agent model configurations that override the default
|
|
# model assignments. Each agent can have a primary model, fallback model, and
|
|
# a chain of additional fallbacks for high availability.
|
|
#
|
|
# For detailed configuration files, see: config/agents/<agent>-models.yaml
|
|
# ==============================================================================
|
|
|
|
agent_model_mappings:
|
|
# -------------------------------------------------------------------------
|
|
# Arbiter - Decision maker / Conflict resolver
|
|
# -------------------------------------------------------------------------
|
|
arbiter:
|
|
primary: anthropic/claude-3-5-sonnet
|
|
fallback: openai/gpt-4-turbo
|
|
fallback_chain:
|
|
- openai/gpt-4o
|
|
- openai/gpt-4
|
|
- anthropic/claude-3-sonnet
|
|
max_tokens: 8192
|
|
temperature: 0.5
|
|
api_key_env: ANTHROPIC_API_KEY
|
|
budget_daily_usd: 25.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Coder - Code generation and review
|
|
# -------------------------------------------------------------------------
|
|
coder:
|
|
primary: anthropic/claude-3-5-sonnet
|
|
fallback: openai/gpt-4o
|
|
fallback_chain:
|
|
- openai/gpt-4-turbo
|
|
- anthropic/claude-3-5-haiku
|
|
- openai/gpt-4o-mini
|
|
max_tokens: 8192
|
|
temperature: 0.7
|
|
api_key_env: ANTHROPIC_API_KEY
|
|
budget_daily_usd: 50.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Steward - Orchestrator
|
|
# -------------------------------------------------------------------------
|
|
steward:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: zai/glm-5-1
|
|
max_tokens: 8192
|
|
temperature: 0.7
|
|
budget_daily_usd: 20.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Alpha - Triad deliberator
|
|
# -------------------------------------------------------------------------
|
|
alpha:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: anthropic/claude-3-5-sonnet
|
|
max_tokens: 8192
|
|
temperature: 0.7
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Beta - Triad deliberator
|
|
# -------------------------------------------------------------------------
|
|
beta:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o
|
|
max_tokens: 8192
|
|
temperature: 0.7
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Charlie - Triad deliberator
|
|
# -------------------------------------------------------------------------
|
|
charlie:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: anthropic/claude-3-5-haiku
|
|
max_tokens: 8192
|
|
temperature: 0.7
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Examiner - Interrogator / Challenger
|
|
# -------------------------------------------------------------------------
|
|
examiner:
|
|
primary: anthropic/claude-3-5-sonnet
|
|
fallback: openai/gpt-4-turbo
|
|
max_tokens: 4096
|
|
temperature: 0.4
|
|
budget_daily_usd: 20.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Explorer - Scout / Opportunity finder
|
|
# -------------------------------------------------------------------------
|
|
explorer:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o-mini
|
|
max_tokens: 4096
|
|
temperature: 0.8
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Sentinel - Guardian / Security
|
|
# -------------------------------------------------------------------------
|
|
sentinel:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: anthropic/claude-3-haiku
|
|
max_tokens: 4096
|
|
temperature: 0.3
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Dreamer - Visionary / Creative
|
|
# -------------------------------------------------------------------------
|
|
dreamer:
|
|
primary: openai/gpt-4-turbo
|
|
fallback: anthropic/claude-3-5-sonnet
|
|
max_tokens: 8192
|
|
temperature: 0.9
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Empath - Diplomat / Emotional intelligence
|
|
# -------------------------------------------------------------------------
|
|
empath:
|
|
primary: anthropic/claude-3-5-sonnet
|
|
fallback: openai/gpt-4o
|
|
max_tokens: 4096
|
|
temperature: 0.8
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Historian - Archivist / Memory keeper
|
|
# -------------------------------------------------------------------------
|
|
historian:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o
|
|
max_tokens: 8192
|
|
temperature: 0.3
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Arbiter - Adjudicator / Decision maker
|
|
# -------------------------------------------------------------------------
|
|
arbiter:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: anthropic/claude-3-5-sonnet
|
|
max_tokens: 8192
|
|
temperature: 0.5
|
|
budget_daily_usd: 25.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Catalyst - Accelerator / Process optimization
|
|
# -------------------------------------------------------------------------
|
|
catalyst:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o-mini
|
|
max_tokens: 4096
|
|
temperature: 0.7
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Chronos - Timekeeper / Schedule management
|
|
# -------------------------------------------------------------------------
|
|
chronos:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o
|
|
max_tokens: 4096
|
|
temperature: 0.5
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Coordinator - Integrator / Cross-agent coordination
|
|
# -------------------------------------------------------------------------
|
|
coordinator:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o
|
|
max_tokens: 4096
|
|
temperature: 0.6
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Echo - Communicator / External communications
|
|
# -------------------------------------------------------------------------
|
|
echo:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o-mini
|
|
max_tokens: 4096
|
|
temperature: 0.7
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Habit Forge - Optimizer / Habit formation
|
|
# -------------------------------------------------------------------------
|
|
habit-forge:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o-mini
|
|
max_tokens: 4096
|
|
temperature: 0.6
|
|
budget_daily_usd: 10.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Metis - Strategist / Strategic planning
|
|
# -------------------------------------------------------------------------
|
|
metis:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: anthropic/claude-3-5-sonnet
|
|
max_tokens: 8192
|
|
temperature: 0.7
|
|
budget_daily_usd: 20.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Nexus - Connector / System integration
|
|
# -------------------------------------------------------------------------
|
|
nexus:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o
|
|
max_tokens: 4096
|
|
temperature: 0.5
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Perceiver - Sensor / Data interpretation
|
|
# -------------------------------------------------------------------------
|
|
perceiver:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: openai/gpt-4o
|
|
max_tokens: 4096
|
|
temperature: 0.6
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Prism - Analyzer / Multi-perspective analysis
|
|
# -------------------------------------------------------------------------
|
|
prism:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: anthropic/claude-3-5-sonnet
|
|
max_tokens: 8192
|
|
temperature: 0.7
|
|
budget_daily_usd: 15.00
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Sentinel Prime - Commander / Security coordination
|
|
# -------------------------------------------------------------------------
|
|
sentinel-prime:
|
|
primary: minimax/MiniMax-M2.7
|
|
fallback: anthropic/claude-3-haiku
|
|
max_tokens: 4096
|
|
temperature: 0.3
|
|
budget_daily_usd: 15.00
|
|
|
|
# ==============================================================================
|
|
# Budget & Rate Limits (Per Agent)
|
|
# ==============================================================================
|
|
budget_settings:
|
|
enabled: true
|
|
|
|
# Per-agent budgets (tokens per day)
|
|
agent_budgets:
|
|
agent/steward: 1000000
|
|
agent/alpha: 500000
|
|
agent/beta: 500000
|
|
agent/charlie: 500000
|
|
agent/examiner: 300000
|
|
agent/explorer: 300000
|
|
agent/sentinel: 300000
|
|
agent/coder: 1000000
|
|
agent/arbiter: 800000
|
|
agent/dreamer: 500000
|
|
agent/empath: 400000
|
|
agent/historian: 500000
|
|
agent/catalyst: 400000
|
|
agent/chronos: 300000
|
|
agent/coordinator: 400000
|
|
agent/echo: 300000
|
|
agent/habit-forge: 300000
|
|
agent/metis: 500000
|
|
agent/nexus: 400000
|
|
agent/perceiver: 400000
|
|
agent/prism: 400000
|
|
agent/sentinel-prime: 500000
|
|
|
|
# Global budget
|
|
global_budget: 5000000
|
|
|
|
# Alert thresholds
|
|
alert_threshold: 0.8
|
|
|
|
# ==============================================================================
|
|
# Observability Settings
|
|
# ==============================================================================
|
|
observability:
|
|
# -------------------------------------------------------------------------
|
|
# Prometheus Metrics
|
|
# -------------------------------------------------------------------------
|
|
prometheus:
|
|
enabled: true
|
|
port: 4000 # Exposed on main LiteLLM port at /metrics
|
|
# Metrics available:
|
|
# - litellm_cost_dollars_total{model, agent}
|
|
# - litellm_tokens_total{model, type=input/output}
|
|
# - litellm_request_count_total{model, status}
|
|
# - litellm_request_latency_seconds{model}
|
|
# - litellm_deployment_failure_responses{model}
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Langfuse Tracing (LLM Observability)
|
|
# -------------------------------------------------------------------------
|
|
langfuse:
|
|
enabled: true
|
|
public_key: os.environ/LANGFUSE_PUBLIC_KEY
|
|
secret_key: os.environ/LANGFUSE_SECRET_KEY
|
|
host: os.environ/LANGFUSE_HOST
|
|
# Trace all LLM calls with costs, latencies, and user sessions
|
|
# Access dashboard at http://localhost:3000
|
|
|
|
# -------------------------------------------------------------------------
|
|
# OpenTelemetry Tracing (Distributed Tracing)
|
|
# -------------------------------------------------------------------------
|
|
opentelemetry:
|
|
enabled: true
|
|
exporter: os.environ/OTEL_EXPORTER_TYPE # otlp, jaeger, zipkin
|
|
endpoint: os.environ/OTEL_EXPORTER_OTLP_ENDPOINT
|
|
|
|
# -------------------------------------------------------------------------
|
|
# LiteLLM WebUI (Built-in Admin Dashboard)
|
|
# -------------------------------------------------------------------------
|
|
# Access at http://localhost:4000/ui
|
|
# Features:
|
|
# - Cost tracking per model/key/user
|
|
# - Token usage analytics
|
|
# - Budget management and alerts
|
|
# - Key management
|
|
# - Spend reports
|
|
ui:
|
|
enabled: true
|
|
access_mode: admin # Requires master key authentication
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Cost Tracking & Budget Alerts
|
|
# -------------------------------------------------------------------------
|
|
cost_tracking:
|
|
enabled: true
|
|
# Track spend by agent passthrough endpoints
|
|
track_agent_costs: true
|
|
# Alert thresholds (percentage of budget)
|
|
alert_threshold_warning: 0.8 # 80% - warning
|
|
alert_threshold_critical: 1.0 # 100% - budget exceeded
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Slack/Webhook Alerts (Optional)
|
|
# -------------------------------------------------------------------------
|
|
slack:
|
|
enabled: false
|
|
webhook_url: os.environ/SLACK_WEBHOOK_URL
|
|
# Alert channels for budget warnings, failures, etc.
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Health Dashboard Integration
|
|
# -------------------------------------------------------------------------
|
|
# Metrics exposed for external dashboards:
|
|
# - GET /metrics (Prometheus format)
|
|
# - GET /spend (JSON spend data)
|
|
# - GET /budget/list (Budget statuses)
|
|
health_dashboard:
|
|
enabled: true
|
|
# Expose metrics for Heretek OpenClaw Health Dashboard
|
|
# Dashboard collects metrics every 30 seconds
|