Struct RunArgs 

pub struct RunArgs {

    pub path: Option<String>,
    pub task: Option<String>,
    pub model: Option<String>,
    pub yolo: bool,
    pub allow: Vec<String>,
    pub max_depth: Option<usize>,
    pub no_seed_commands: bool,
    pub workdir: Option<PathBuf>,
    pub json: bool,
    pub count: usize,
    pub output_format: Option<String>,
    pub output_instructions: Option<String>,
    pub output_schema: Option<String>,
    pub regions: HashMap<String, String>,
}

Arguments for lev run.

Fields

path: Option<String>

Path to the agent (a manifest file, its directory, or an installed name).

task: Option<String>

Task prompt, or the path of a file holding it. Left off, your editor opens on a template for you to write it in.

model: Option<String>

Model override (provider/model or a bare model name).

yolo: bool

Run unattended: approve every tool call, and answer the agent’s own prompts (ask_user_*, interaction points) instead of waiting for a person.

One exception, and it is the one that looks like a hang: an interaction point declaring unattended = "ask" still holds for a person. The bundled coder’s plan approval can, deliberately, because everything after it writes code. Such a run parks in Waiting until [limits] interaction_timeout_secs (default 3600) releases it. Lower that to bound the wait.

allow: Vec<String>

Allow a tool outright (repeatable).

max_depth: Option<usize>

Override the blueprint’s max sub-agent tree depth.

no_seed_commands: bool

Refuse the blueprint’s seed = { command = "..." } regions. Those run a shell command at spawn - before the first inference, and so before any approval prompt. See lev validate <path> to inspect them first.

workdir: Option<PathBuf>

Working directory for the run (default: the directory lev run is invoked from). The agent’s file tools are confined to it, and relative [read_paths] entries resolve against it.

json: bool

Print the spawned run as JSON instead of a sentence, for a caller that has to parse the run id back out and poll lev ps --json. With --count above 1 the JSON is an array, one object per run.

count: usize

Start this many runs of the same agent and task, each under its own run id, from one invocation. One process launch and one socket dial per run caps a shell loop near 60 spawns/second; the daemon itself has no run cap, and a single invocation carrying the batch spawns as fast as the daemon accepts.

output_format: Option<String>

Ask for the final output in a particular shape, overriding whatever the blueprint declares. Any label works - markdown, json, xml, a2ui, a media type, a house format - because nothing converts between shapes: the label and any instructions are handed to the model, which produces the bytes. Read the answer back with lev result <run-id>.

Naming a format without --output-schema drops a schema the blueprint declared, since a check written for one shape says nothing about another.

output_instructions: Option<String>

Extra guidance about the shape, passed to the model alongside --output-format. This is how an unusual format gets explained.

output_schema: Option<String>

A JSON Schema (inline, or @path to a file) the final output must satisfy. The only thing that ever inspects the answer’s contents, and it only happens because you asked: a submission that fails is refused back to the agent to correct.

regions: HashMap<String, String>

Dynamic per-region seed flags (--<region> <text|@file>), collected by an argv pre-scan in the binary since region names are blueprint-defined. clap skips this field; it is populated after parsing.

Trait Implementations

impl Args for RunArgs

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 Clone for RunArgs

fn clone(&self) -> RunArgs

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RunArgs

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

Formats the value using the given formatter.

impl Default for RunArgs

fn default() -> RunArgs

Returns the “default value” for a type.

impl FromArgMatches for RunArgs

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.