Skip to main content

Configuration Guide

Configure Memfit AI to suit your needs.

AI Provider Settings​

Supported Providers​

ProviderModelsFeatures
OpenAIGPT-4, GPT-3.5-turboBest reasoning
AnthropicClaude 3, Claude 2Long context
LocalOllama, LMStudioPrivacy
CustomAny OpenAI-compatibleFlexibility

Configuration​

ai:
  provider: openai
  model: gpt-4
  api_key: sk-xxxxx
  base_url: https://api.openai.com/v1  # Optional
  temperature: 0.7
  max_tokens: 4096

Using Proxy​

For network restrictions:

ai:
  provider: openai
  model: gpt-4
  proxy: http://127.0.0.1:7890

Execution Settings​

ReAct Loop​

react:
  max_steps: 20          # Maximum iterations
  spin_threshold: 3       # Steps before spin detection
  reflection_on_error: true
  timeout: 300            # Seconds

Plan Engine​

plan:
  max_depth: 5            # Maximum task tree depth
  parallel_execution: true
  require_approval: true   # Human review required

Memory Configuration​

Short-term Memory​

memory:
  short_term:
    max_messages: 50
    include_system: true

Long-term Memory​

memory:
  long_term:
    enabled: true
    score_threshold: 0.6
    max_entries: 10000
    vector_db: sqlite       # or postgres, qdrant

RAG Configuration​

rag:
  enabled: true
  top_k: 5
  min_similarity: 0.7
  sources:
    - type: local
      path: /path/to/docs
    - type: web
      url: https://docs.example.com

Tool Configuration​

Enabling Tools​

tools:
  file_system:
    enabled: true
    allowed_paths:
      - /home/user/projects
      - /tmp
    
  shell:
    enabled: true
    require_confirmation: true
    
  network:
    enabled: true
    allowed_hosts:
      - "*.example.com"
      - "192.168.1.*"

Tool Permissions​

permissions:
  read: true
  write: true        # Requires confirmation
  execute: true      # Requires confirmation
  network: true
  sensitive: confirm  # always, never, confirm

Forge Configuration​

Loading Forges​

forges:
  - name: java-audit
    enabled: true
    priority: high
    
  - name: web-security
    enabled: true
    
  - name: custom-forge
    path: /path/to/forge

Yakit Integration​

Engine Connection​

yakit:
  engine_port: 8087
  auto_start: true
  log_level: info

UI Preferences​

yakit:
  ui:
    theme: dark
    language: en       # or zh-Hans
    show_traces: true
    auto_scroll: true

Environment Variables​

Override configuration with environment variables:

export MEMFIT_AI_PROVIDER=openai
export MEMFIT_AI_MODEL=gpt-4
export MEMFIT_AI_API_KEY=sk-xxxxx
export MEMFIT_MAX_STEPS=30

Configuration File Location​

Default locations:

PlatformPath
macOS~/.config/memfit/config.yaml
Linux~/.config/memfit/config.yaml
Windows%APPDATA%\memfit\config.yaml

Example Complete Configuration​

ai:
  provider: openai
  model: gpt-4
  api_key: ${OPENAI_API_KEY}
  temperature: 0.7

react:
  max_steps: 25
  spin_threshold: 3

plan:
  max_depth: 5
  require_approval: true

memory:
  long_term:
    enabled: true
    score_threshold: 0.65

tools:
  file_system:
    enabled: true
  shell:
    enabled: true
    require_confirmation: true
  network:
    enabled: true

forges:
  - name: web-security
    enabled: true