Struct TaskNode 

pub struct TaskNode {

    pub id: TaskId,
    pub title: String,
    pub description: String,
    pub agent_hint: Option<String>,
    pub status: TaskStatus,
    pub depends_on: Vec<TaskId>,
    pub result: Option<TaskResult>,
    pub assigned_agent: Option<String>,
    pub retry_count: u32,
    pub predicate_rerun_count: u32,
    pub failure_strategy: Option<FailureStrategy>,
    pub max_retries: Option<u32>,
    pub execution_mode: ExecutionMode,
    pub verify_predicate: Option<VerifyPredicate>,
    pub predicate_outcome: Option<PredicateOutcome>,
}

A single node in the task DAG.

Constructed by Planner and stored inside a TaskGraph. The scheduler drives each node through its TaskStatus lifecycle.

Examples

use zeph_orchestration::{TaskNode, TaskStatus, ExecutionMode};

let node = TaskNode::new(0, "fetch data", "Download the dataset from source.");
assert_eq!(node.status, TaskStatus::Pending);
assert!(node.depends_on.is_empty());
assert_eq!(node.execution_mode, ExecutionMode::Parallel);

Fields

id: TaskId

Dense zero-based index. Invariant: tasks[i].id == TaskId(i).

title: String

Short, human-readable task title.

description: String

Full task description passed verbatim to the assigned sub-agent as its prompt.

agent_hint: Option<String>

Preferred agent name suggested by the planner; None lets the router decide.

status: TaskStatus

Current lifecycle status.

depends_on: Vec<TaskId>

Indices of tasks this node depends on.

result: Option<TaskResult>

Result populated by the scheduler after the sub-agent finishes.

assigned_agent: Option<String>

Agent name actually assigned by the router at dispatch time.

retry_count: u32

Number of times this task has been retried so far (execution retries only).

predicate_rerun_count: u32

Number of predicate-driven re-runs for this task (independent of retry_count).

failure_strategy: Option<FailureStrategy>

Per-task failure strategy override; None means use TaskGraph::default_failure_strategy.

max_retries: Option<u32>

Maximum retry attempts for this task; None means use TaskGraph::default_max_retries.

execution_mode: ExecutionMode

LLM planner annotation. Old SQLite-stored JSON without this field deserialises to the default (Parallel).

verify_predicate: Option<VerifyPredicate>

Per-subtask verification predicate (predicate gate).

When Some, the task’s output must satisfy this criterion before downstream tasks may consume it. The scheduler emits SchedulerAction::VerifyPredicate after task completion and blocks downstream dispatch until predicate_outcome.is_some().

predicate_outcome: Option<PredicateOutcome>

Outcome of the most recent predicate evaluation.

None means the gate has not been evaluated yet (in-memory only; restart re-evaluates any pending predicates). The scheduler re-emits VerifyPredicate on every tick while this is None and verify_predicate.is_some().

Implementations

impl TaskNode

pub fn new( id: u32, title: impl Into<String>, description: impl Into<String>, ) -> Self

Create a new pending task with the given index.

Trait Implementations

impl Clone for TaskNode

fn clone(&self) -> TaskNode

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TaskNode

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TaskNode

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for TaskNode

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

where __S: Serializer,

Serialize this value into the given Serde serializer.