Struct RunListEntry 

pub struct RunListEntry {
    pub run_id: String,
    pub status: AgentStatus,
    pub wait_reason: Option<WaitReason>,
    pub stage: String,
    pub stage_index: Option<usize>,
    pub num_stages: Option<usize>,
    pub iteration: usize,
    pub tool_calls: usize,
    pub last_progress_at: Option<i64>,
    pub unattended: bool,
    pub empty_output: bool,
    pub read_paths: Option<ReadPathGrantCounts>,
}

One row of a run listing (ControlRequest::List): a live run, its status, and enough context to judge whether that status is a problem.

lev ps used to be a run id and a status word, which is why issue #184 happened: waiting on its own says nothing about whether a person is needed, and there was no way to tell a run that had moved a second ago from one that had been stopped for an hour. Everything here is read straight off the live world, so it is the daemon’s own view, not a re-read of meta.json.

Fields

run_id: String

The run id (lev ps’s first column, and what lev kill takes).

status: AgentStatus

The agent’s live status.

wait_reason: Option<WaitReason>

Why the status is AgentStatus::Waiting; None for every other status, and for a Waiting the host cannot attribute.

stage: String

The stage the agent is in.

stage_index: Option<usize>

Zero-based index of that stage, when the agent tracks one.

num_stages: Option<usize>

How many stages the blueprint has.

iteration: usize

Iterations completed in the current stage.

tool_calls: usize

Cumulative tool calls across the run.

last_progress_at: Option<i64>

Unix seconds when this run last actually moved (see PersistWatermark). Distinct from meta.json’s updated_at, which also advances on a heartbeat and so cannot be used to tell a working run from a wedged one.

unattended: bool

Whether this run is unattended (--yolo). An unattended run should never be sitting on a prompt; if it is, something dropped the flag.

empty_output: bool

Whether this run finished having modified nothing, when its blueprint gave it a way to. Only ever true for a run that has stopped.

The flag itself is as old as issue #107, but nothing ever showed it: it went into meta.json and was read back only on restart, so a run that finished with no work to show for it looked exactly like one that succeeded. Defaulted for the same reason as unattended - an older daemon simply omits it.

read_paths: Option<ReadPathGrantCounts>

How much of this run’s [read_paths] its config granted at spawn. None for a blueprint that declares none, which is nearly every agent.

Worth a column of its own because an ungranted declaration is inert: the run is up, looks healthy, and will be refused the reads its author designed it around.

Trait Implementations

impl Clone for RunListEntry

fn clone(&self) -> RunListEntry

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RunListEntry

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for RunListEntry

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for RunListEntry

fn eq(&self, other: &RunListEntry) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for RunListEntry

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for RunListEntry