Struct Cli 

pub struct Cli {

    pub backend: BackendKind,
    pub lock: PathBuf,
    pub tcp: Option<String>,
    pub uds: Option<PathBuf>,
    pub pipe: Option<String>,
    pub group: Option<String>,
    pub active_permits: usize,
    pub queue_depth: usize,
    pub ready_timeout_secs: u64,
    pub model_path: Option<PathBuf>,
    pub model_sha256: Option<String>,
    pub n_ctx: u32,
    pub n_gpu_layers: i32,
    pub openai_base_url: Option<String>,
    pub openai_api_key: Option<String>,
    pub openai_model: Option<String>,
    pub openai_timeout_secs: u64,
    pub bedrock_region: Option<String>,
    pub bedrock_model_id: Option<String>,
    pub bedrock_bearer_token: Option<String>,
    pub bedrock_endpoint: Option<String>,
    pub bedrock_timeout_secs: u64,
    pub api_key: Option<String>,
    pub config: Option<PathBuf>,
    pub admin_addr: Option<PathBuf>,
    pub v2: bool,
    pub v2_addr: Option<PathBuf>,
    pub v2_tcp: Option<String>,
    pub embed: bool,
    pub embed_addr: Option<PathBuf>,
    pub embed_tcp: Option<String>,
}

Top-level CLI for inferd-daemon.

Fields

backend: BackendKind

Backend to load at startup.

lock: PathBuf

Path to the single-instance lock file. The lock is held for the lifetime of the daemon process.

tcp: Option<String>

Loopback TCP bind address. Mutually exclusive with --uds and --pipe.

uds: Option<PathBuf>

Unix domain socket path. Mutually exclusive with --tcp and --pipe. Unix only.

pipe: Option<String>

Windows named pipe path (e.g. \\.\pipe\inferd-infer). Mutually exclusive with --tcp and --uds. Windows only.

group: Option<String>

Group name for the UDS (Unix only). Ignored on other transports.

active_permits: usize

Active generations served concurrently. v0.1 invariant is 1; values above 1 are reserved for v0.2 continuous-batching backends.

queue_depth: usize

Maximum waiting queue depth. Submits beyond this return code: queue_full immediately.

ready_timeout_secs: u64

Seconds to wait for the backend to report ready before failing startup.

model_path: Option<PathBuf>

Path to the GGUF model file. Required when --backend llamacpp.

model_sha256: Option<String>

Optional expected SHA-256 of the model file as a hex string (64 chars). When present, the daemon verifies the file before loading via subtle::ConstantTimeEq (THREAT_MODEL F-5).

n_ctx: u32

Llama.cpp context window in tokens. Default 8192.

n_gpu_layers: i32

Llama.cpp GPU layer offload count. 0 = CPU-only. GPU support requires the cuda/metal/vulkan/rocm cargo feature at build time.

openai_base_url: Option<String>

Base URL of the upstream OpenAI-compat endpoint, no trailing slash and no path (the adapter appends /v1/chat/completions). Required when --backend openai-compat. Examples: https://api.openai.com, http://localhost:11434, https://openrouter.ai.

openai_api_key: Option<String>

Bearer token for the OpenAI-compat upstream. Sent as Authorization: Bearer <value>. Pass an empty string to skip the header entirely for self-hosted endpoints. Resolves from --openai-api-key, then INFERD_OPENAI_API_KEY, then OPENAI_API_KEY (the de-facto env name most providers’ SDKs already use).

openai_model: Option<String>

Upstream model identifier echoed in the request model field — provider-specific (e.g. gpt-4o-mini, llama3.1:8b, meta-llama/Meta-Llama-3-70B-Instruct). Required when --backend openai-compat.

openai_timeout_secs: u64

Total request timeout for OpenAI-compat calls, in seconds. Default 300 (5 minutes) — long enough for a slow first-token from a cold cloud model, short enough to surface stuck requests rather than hang forever.

bedrock_region: Option<String>

AWS region the Bedrock endpoint lives in, e.g. us-east-1, eu-central-1. Required when --backend bedrock-invoke. Used for both the endpoint host and SigV4 signing scope.

bedrock_model_id: Option<String>

Bedrock model id (URL-encoded by the adapter), e.g. anthropic.claude-3-5-sonnet-20241022-v2:0. Required when --backend bedrock-invoke.

bedrock_bearer_token: Option<String>

Pre-issued Bedrock bearer token (AWS_BEARER_TOKEN_BEDROCK shape, AWS rolled this out in 2025-06). When set, the adapter sends Authorization: Bearer <value> and skips SigV4. When unset, the adapter falls back to the standard AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY (+ optional AWS_SESSION_TOKEN) chain via SigV4 signing.

bedrock_endpoint: Option<String>

Override the Bedrock endpoint host. Empty/absent → default bedrock-runtime.<region>.amazonaws.com. Useful for VPC endpoints / integration tests.

bedrock_timeout_secs: u64

Total request timeout for Bedrock calls, in seconds. Default 300 (5 minutes).

api_key: Option<String>

Optional pre-shared API key. When set, TCP clients MUST send {"type":"auth","key":"<this value>"} as their first NDJSON frame on the connection or the daemon closes the connection. UDS and named-pipe transports ignore this — kernel-attested peer credentials (F-7) do the work there.

Comparison is constant-time. THREAT_MODEL F-8.

config: Option<PathBuf>

Path to the operator JSON config file. Default ~/.inferd/config.json. When present, fetch + auto-pull are driven from it; CLI flags (--model-path, --model-sha256, --n-ctx, --n-gpu-layers) override config-file values when both are supplied. When absent, the daemon falls back to CLI-flag-only operation (dev mode).

admin_addr: Option<PathBuf>

Admin endpoint path. Defaults per-platform to the path documented in docs/protocol-v1.md §“Admin endpoint” — e.g. /run/inferd/admin.sock on Linux, \\.\pipe\inferd-admin on Windows. Override for tests / non-default deployments.

v2: bool

Enable the v2 inference endpoint per ADR 0015. v2 binds on a separate socket from v1: infer.v2.sock on Unix / \\.\pipe\inferd-infer-v2 on Windows. v1 stays on its own socket and is unaffected.

Phase 1B: the v2 endpoint accepts and validates v2 requests but returns Error{code:internal, message:"v2 generation not implemented"} because the Backend trait does not yet expose generate_v2. Use this to integration-test middleware that will speak v2 once Phase 2A lands.

v2_addr: Option<PathBuf>

Override the default v2 inference endpoint path. Mirrors --uds / --pipe for v2; on Linux/macOS this is a UDS path, on Windows a named-pipe path. Has no effect unless --v2 is also set.

v2_tcp: Option<String>

Loopback TCP bind address for the v2 endpoint. Mutually exclusive with --v2-addr. Useful for tests that don’t want the platform default (UDS / named pipe). Has no effect unless --v2 is also set.

embed: bool

Enable the embed inference endpoint per ADR 0017. The embed endpoint binds on a separate socket from v1/v2: infer.embed.sock on Unix / \\.\pipe\inferd-infer-embed on Windows. Has no effect unless the active backend’s capabilities().embed is true (capability-driven binding).

embed_addr: Option<PathBuf>

Override the default embed inference endpoint path. Mirrors --uds / --pipe for embed; on Linux/macOS this is a UDS path, on Windows a named-pipe path. Has no effect unless --embed is also set.

embed_tcp: Option<String>

Loopback TCP bind address for the embed endpoint. Mutually exclusive with --embed-addr. Has no effect unless --embed is also set.

Implementations

impl Cli

pub fn require_one_transport(&self) -> Result<(), &'static str>

Validate that exactly one transport is selected. clap enforces mutual exclusion; this checks the at-least-one part.

Trait Implementations

impl Args for Cli

fn group_id() -> Option<Id>

Report the ArgGroup::id for this set of arguments

fn augment_args<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate Self via FromArgMatches::from_arg_matches_mut

fn augment_args_for_update<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate self via FromArgMatches::update_from_arg_matches_mut

impl CommandFactory for Cli

fn command<'b>() -> Command

Build a Command that can instantiate Self.

fn command_for_update<'b>() -> Command

Build a Command that can update self.

impl Debug for Cli

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

Formats the value using the given formatter.

impl FromArgMatches for Cli

fn from_arg_matches(__clap_arg_matches: &ArgMatches) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn from_arg_matches_mut( __clap_arg_matches: &mut ArgMatches, ) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn update_from_arg_matches( &mut self, __clap_arg_matches: &ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

fn update_from_arg_matches_mut( &mut self, __clap_arg_matches: &mut ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

impl Parser for Cli

fn parse() -> Self

Parse from std::env::args_os(), exit on error.

fn try_parse() -> Result<Self, Error>

Parse from std::env::args_os(), return Err on error.

fn parse_from<I, T>(itr: I) -> Self

where I: IntoIterator<Item = T>, T: Into<OsString> + Clone,

Parse from iterator, exit on error.

fn try_parse_from<I, T>(itr: I) -> Result<Self, Error>

where I: IntoIterator<Item = T>, T: Into<OsString> + Clone,

Parse from iterator, return Err on error.

fn update_from<I, T>(&mut self, itr: I)

where I: IntoIterator<Item = T>, T: Into<OsString> + Clone,

Update from iterator, exit on error.

fn try_update_from<I, T>(&mut self, itr: I) -> Result<(), Error>

where I: IntoIterator<Item = T>, T: Into<OsString> + Clone,

Update from iterator, return Err on error.