Application Configuration¶

application.conf is the main configuration file, located in the project conf/ directory.

LLM Configuration¶

General Settings¶

Key	Description	Allowed Values
`llm.choice-provider`	Default AI provider	`openai` / `anthropic`

Capability Declaration¶

Tells the framework which features the selected provider supports for each model tier. Top-level llm.capability.* switches are the default capabilities; lite, flash, and pro override the corresponding tier.

Key	Description
`llm.capability.completion`	Supports text completion
`llm.capability.prompt-caching`	Supports prompt caching
`llm.capability.temperature`	Supports temperature sampling
`llm.capability.tools`	Supports function calling / tools
`llm.capability.tool-choice`	Supports forced tool choice
`llm.capability.multiple-choices`	Supports multiple candidates
`llm.capability.thinking`	Supports thinking / reasoning mode
`llm.capability.vision-image`	Supports image input
`llm.capability.lite.*`	Capability overrides for the lite tier
`llm.capability.flash.*`	Capability overrides for the flash tier
`llm.capability.pro.*`	Capability overrides for the pro tier

Parameter Overrides¶

llm.param.lite, llm.param.flash, and llm.param.pro can be used to override default LLM parameters (such as temperature, maxTokens, etc.) for each model tier.

Providers¶

llm {
  choice-provider: "openai"

  providers {
    openai {
      api-key: ${?OPENAI_API_KEY}
      base-url: "https://api.openai.com/v1"
      models {
        lite: "gpt-4o-mini"
        flash: "gpt-4o"
        pro: "gpt-4.1"
      }
    }
  }
}

Models are organized in three tiers: lite, flash, pro. choice-provider sets the default provider.

Each provider supports the same set of keys: api-key, base-url, and models.{lite,flash,pro}.

Usage Pricing¶

Key	Description
`llm.usage-pricing.price-unit`	Pricing unit, e.g. `USD`
`llm.usage-pricing.lite.input-cache-hit`	Lite cached input price
`llm.usage-pricing.lite.input-cache-miss`	Lite uncached input price
`llm.usage-pricing.lite.output`	Lite output price
`llm.usage-pricing.flash.*`	Flash tier prices
`llm.usage-pricing.pro.*`	Pro tier prices

State Tuning¶

Emotion¶

Key	Description
`state-tuning.emotion.emotion-retention-high`	High emotion retention rate
`state-tuning.emotion.emotion-retention-medium`	Medium emotion retention rate
`state-tuning.emotion.emotion-retention-low`	Low emotion retention rate
`state-tuning.emotion.mood-retention`	Mood retention rate
`state-tuning.emotion.mood-emotion-influence`	Mood influence on emotion
`state-tuning.emotion.emotion-half-life-seconds`	Emotion half-life in seconds
`state-tuning.emotion.mood-half-life-seconds`	Mood half-life in seconds

Flow¶

Key	Description
`state-tuning.flow.min-ratio-of-desire`	Minimum desire ratio
`state-tuning.flow.min-value-min`	Minimum value lower bound
`state-tuning.flow.min-value-max`	Minimum value upper bound
`state-tuning.flow.decay-normal-per-minute`	Normal decay per minute
`state-tuning.flow.decay-negative-per-minute`	Negative decay per minute
`state-tuning.flow.core-interest-base-charge`	Core interest base charge
`state-tuning.flow.continuous-interaction-base-charge`	Continuous interaction base charge
`state-tuning.flow.deep-reply-base-charge`	Deep reply base charge
`state-tuning.flow.group-resonance-base-charge`	Group resonance base charge
`state-tuning.flow.negative-penalty`	Negative penalty
`state-tuning.flow.topic-interrupt-penalty`	Topic interrupt penalty
`state-tuning.flow.repeat-topic-penalty`	Repeat topic penalty
`state-tuning.flow.low-activity-penalty`	Low activity penalty
`state-tuning.flow.getting-better-threshold`	Getting-better threshold
`state-tuning.flow.burst-threshold`	Burst threshold

Volition¶

Key	Description
`state-tuning.volition.base-desire-default`	Default base desire
`state-tuning.volition.keyword-hit-max-stimulus`	Maximum keyword-hit stimulus
`state-tuning.volition.busy-group-stimulus`	Busy group stimulus
`state-tuning.volition.indirect-mention-stimulus`	Indirect mention stimulus
`state-tuning.volition.emotional-resonance-stimulus`	Emotional resonance stimulus
`state-tuning.volition.reset-stimulus-amount`	Reset stimulus amount
`state-tuning.volition.fatigue-on-speak`	Fatigue on speaking
`state-tuning.volition.fatigue-decay-low-arousal`	Fatigue decay at low arousal
`state-tuning.volition.fatigue-decay-normal`	Normal fatigue decay
`state-tuning.volition.stimulus-decay-normal`	Normal stimulus decay
`state-tuning.volition.stimulus-decay-high-flow`	Stimulus decay at high flow
`state-tuning.volition.normal-speak-threshold`	Normal speak threshold
`state-tuning.volition.high-flow-speak-threshold`	High-flow speak threshold
`state-tuning.volition.high-flow-threshold`	High-flow threshold
`state-tuning.volition.arousal-impulse-weight`	Arousal impulse weight
`state-tuning.volition.negative-pleasure-penalty-weight`	Negative pleasure penalty weight
`state-tuning.volition.flow-bonus-start`	Flow bonus start value
`state-tuning.volition.flow-bonus-weight`	Flow bonus weight

Memory¶

Key	Description
`state-tuning.memory.batch-limit`	Max messages per batch
`state-tuning.memory.min-messages`	Minimum messages to trigger memory
`state-tuning.memory.stale-recall-days`	Stale recall days

Summary¶

Key	Description
`state-tuning.summary.batch-limit`	Max messages per batch
`state-tuning.summary.min-messages`	Minimum messages to trigger summary
`state-tuning.summary.retention-days`	Summary retention days

Meme¶

Key	Description
`state-tuning.meme.analyze-threshold`	Required seen count to analyze
`state-tuning.meme.max-contexts`	Max contexts to keep
`state-tuning.meme.cleanup-days`	Cleanup unused memes after days
`state-tuning.meme.low-heat-seen-threshold`	Low-heat seen threshold

Evolution¶

Key	Description
`state-tuning.evolution.default-weight`	Default weight
`state-tuning.evolution.active-weight-threshold`	Active weight threshold
`state-tuning.evolution.min-weight-threshold`	Minimum weight threshold
`state-tuning.evolution.increase-on-use`	Weight increase on use
`state-tuning.evolution.decay-per-cycle`	Decay per cycle
`state-tuning.evolution.decrease-on-negative`	Weight decrease on negative feedback
`state-tuning.evolution.stale-days`	Stale days
`state-tuning.evolution.recent-range-hours`	Recent range in hours
`state-tuning.evolution.recent-message-limit`	Recent message limit
`state-tuning.evolution.active-limit`	Active word limit

OneBot Connection¶

Multiple bots can be configured under onebot.bots, each as a named block.

onebot {
  bots {
    erii {
      ws: "ws://127.0.0.1:3001"
      token: ${?NAPCAT_TOKEN}
      role-id: "erii"
      groups {
        123456789 {
          admins: []
          desire: 15
        }
      }
      enabled-plugins: []
      disabled-plugins: []
    }
  }
}

Bot Settings¶

Key	Description
`onebot.bots.<name>.ws`	Bot WebSocket address
`onebot.bots.<name>.token`	Bot connection token
`onebot.bots.<name>.role-id`	Bot persona ID

Group Settings¶

Key	Description
`onebot.bots.<name>.groups.<id>.admins`	Group admin QQ numbers
`onebot.bots.<name>.groups.<id>.desire`	Group activity / desire level

Group Overrides¶

Key	Description
`onebot.bots.<name>.groups-override.enable-groups`	Override global group whitelist
`onebot.bots.<name>.groups-override.debug-group-id`	Override global debug group
`onebot.bots.<name>.groups-override.message-redirect-map`	Message redirect map, format `source:target`

Plugin Control¶

Key	Description
`onebot.bots.<name>.enabled-plugins`	Whitelist only specified plugins
`onebot.bots.<name>.disabled-plugins`	Blacklist specified plugins

When both are present, the whitelist takes precedence.

Group Configuration¶

groups {
  enable-groups: []
  debug-group-id: null
  message-redirect-map: []
}

Key	Description
`groups.enable-groups`	Globally enabled group IDs
`groups.debug-group-id`	Debug group ID
`groups.message-redirect-map`	Message redirect rules, format `source:target`

Other Components¶

embedding {
  provider: "bytedance"
  api-key: ${?EMBEDDING_API_KEY}
  url: "https://ark.cn-beijing.volces.com/api/v3/embeddings/multimodal"
  model: "doubao-embedding-vision-251215"
}

search {
  provider: "exa"
  api-key: ${?SEARCH_API_KEY}
  url: "https://api.exa.ai/search"
}

vision {
  provider: "minimax"
  api-key: ${?VISION_API_KEY}
  url: "https://api.minimaxi.com/v1/coding_plan/vlm"
}

browser {
  download: false
  playwright-url: ${?PLAYWRIGHT_HOST}
  external-host: ${?EXTERNAL_HOST}
}

proxy {
  http: ${?HTTP_PROXY}
  socks: ${?SOCKS_PROXY}
}

Security¶

Configure the admin page username and password via environment variables. See Environment Variables.

Key	Description
`security.username`	Admin username
`security.password`	Admin password