Struct Settings 

pub struct Settings {

    pub allow_cloud_ai: bool,
    pub allow_local_ai: bool,
    pub local_ai_base_url: Option<String>,
    pub llama_server_binary: Option<String>,
    pub llama_server_model: Option<String>,
    pub llama_params: HashMap<String, LlamaParams>,
    pub local_ai_models: Vec<String>,
    pub providers_enabled: HashMap<String, bool>,
    pub theme: String,
    pub confirm_destructive: bool,
    pub scheduler_enabled: bool,
    pub scheduler_poll_secs: u64,
    pub max_agentic_iterations: u32,
    pub max_agentic_seconds: u32,
    pub max_agentic_tokens: u32,
}

Fields

allow_cloud_ai: bool

Master toggle for cloud-AI nodes. Defaults to false - Flow’s posture is zero-egress local inference. Cloud AI is an explicit, opt-in carve-out (see docs/adr/0004-cloud-ai-carve-out.md).

allow_local_ai: bool

Master toggle for the local OpenAI-compatible provider (on-device inference server: Ollama / LM Studio / llama.cpp). Defaults to true - a localhost call is not network egress, so it stays usable even with allow_cloud_ai off. The local provider is still subject to the per-provider providers_enabled map.

local_ai_base_url: Option<String>

Base address of the local OpenAI-compatible server. Stored as an origin (e.g. http://127.0.0.1:1234); the runtime appends /v1/chat/completions (and /v1/models for discovery). A fully qualified URL is also accepted and normalised. Defaults to LM Studio’s address.

llama_server_binary: Option<String>

Path to the llama-server (llama.cpp) binary flow-studio launches to host an LLM locally. None until the user configures it (or a bundled sidecar is added in a later phase).

llama_server_model: Option<String>

Path to the LLM the managed server should load. Set by the Model Hub when a downloaded LLM is selected as active.

llama_params: HashMap<String, LlamaParams>

Per-model llama-server load parameters (the Model Hub “Load settings” panel), keyed by catalog model id. Each model remembers its own config.

local_ai_models: Vec<String>

Model ids discovered from the local server’s /v1/models, cached so the node inspector’s model dropdown can offer them without a live probe. Populated by the Settings “Test connection” action; empty until then (the provider’s static default is used as a fallback).

providers_enabled: HashMap<String, bool>

Per-provider enable map. Even when allow_cloud_ai is true, individual providers can be disabled here.

theme: String

UI theme. "dark", "light", or "dark-projector" - passed to Precision’s setTheme at startup and on toggle. Defaults to "dark" to match Flow Studio’s design language.

confirm_destructive: bool

Per-step destructive-action confirmation gate (roadmap E1). When on, a run pauses before any node that performs a destructive operation (file delete, rm, git push, …) and asks the user to confirm or cancel. Defaults to true - a safe default the user can opt out of.

scheduler_enabled: bool

Master switch for the background scheduler. Defaults on so persisted timers keep firing unless the user explicitly pauses scheduling.

scheduler_poll_secs: u64

Poll interval in seconds for desktop/server scheduler loops.

max_agentic_iterations: u32

Safety cap on auto-accepted fixes in one autonomous run - the single budget shared by the backend convergence loop (agentic_run_loop) and the frontend monitor auto-fix loop, so neither path hardcodes its own (roadmap E1). Defaults to crate::MAX_AGENTIC_ITERATIONS.

max_agentic_seconds: u32

Wall-clock ceiling (seconds) for one autonomous convergence run (agentic_run_loop); 0 = unlimited (only the iteration cap bounds the loop). The loop stops starting new iterations once this elapses (roadmap E1 autonomous run budgets).

max_agentic_tokens: u32

Cumulative token ceiling for one autonomous convergence run - the sum of input+output tokens across the executed flows’ AI nodes; 0 = unlimited. The loop stops starting new iterations once it’s exceeded (roadmap E1 autonomous run budgets).

Trait Implementations

impl Clone for Settings

fn clone(&self) -> Settings

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Settings

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for Settings

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for Settings

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Settings

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.