Struct Agent 

pub struct Agent {

    pub id: String,
    pub name: String,
    pub harness: Harness,
    pub state: AgentState,
    pub activity: Activity,
    pub pid: Option<u32>,
    pub session_id: Option<String>,
    pub session_path: Option<PathBuf>,
    pub cwd: Option<PathBuf>,
    pub model: Option<String>,
    pub harness_version: Option<String>,
    pub usage: TokenUsage,
    pub cost_usd: f64,
    pub cost_breakdown: CostBreakdown,
    pub price_source: Option<PriceSource>,
    pub unpriced_tokens: u64,
    pub turns: u64,
    pub subagent_turns: u64,
    pub tool_calls: u64,
    pub web_searches: u64,
    pub spans: Vec<ToolSpan>,
    pub age_secs: u64,
    pub idle_secs: Option<u64>,
    pub cpu_percent: f32,
    pub rss_bytes: u64,
    pub process_count: usize,
    pub mcp_count: usize,
    pub mcp_servers: Vec<McpServer>,
    pub tree: Option<ProcNode>,
    pub attribution: Attribution,
    pub shares_process: bool,
    pub parse_warning: Option<String>,
}

A single row in the agent table.

Fields

id: String

Stable identity across refreshes: pid:<n> for live agents, session:<id> for stopped ones.

name: String

harness: Harness

state: AgentState

activity: Activity

pid: Option<u32>

session_id: Option<String>

session_path: Option<PathBuf>

cwd: Option<PathBuf>

model: Option<String>

harness_version: Option<String>

usage: TokenUsage

cost_usd: f64

USD spent on messages whose model had a known price.

cost_breakdown: CostBreakdown

cost_usd by kind of token, so a figure that differs from another tool’s can be traced to the one line that differs.

price_source: Option<PriceSource>

Where the price of this row’s model came from; None when it has none.

unpriced_tokens: u64

Tokens on messages whose model had no known price (so cost_usd is a floor).

turns: u64

subagent_turns: u64

tool_calls: u64

web_searches: u64

Server-side web searches the model ran, billed per search on top of tokens. Counted for every harness; priced only where the price table has a rate (Anthropic’s, for Claude Code).

spans: Vec<ToolSpan>

The most recent spans, oldest first: tool calls, inferences and turns. Bounded; see harness::MAX_SPANS.

age_secs: u64

Seconds since the process started (live) or since the last transcript write (stopped).

idle_secs: Option<u64>

Seconds since the transcript was last written.

cpu_percent: f32

rss_bytes: u64

process_count: usize

mcp_count: usize

mcp_servers: Vec<McpServer>

The MCP servers this agent uses, one row each, from the process tree and the transcript. See McpServer.

tree: Option<ProcNode>

attribution: Attribution

How the session was attributed to the process, for debugging attribution.

shares_process: bool

True when another row shares this pid and carries its CPU, memory and process counts. One Codex app-server hosts many conversations, so its threads each get a row while the process is only counted once. Absent in snapshots written before 0.2.0.

parse_warning: Option<String>

Set when the transcript parsed but its usage records did not, which means this row’s tokens and cost are not to be believed. Almost always a harness that changed its format under us.

Trait Implementations

impl Clone for Agent

fn clone(&self) -> Agent

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Agent

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Agent

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Agent

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

where __S: Serializer,

Serialize this value into the given Serde serializer.