Enum AgentMessage 

pub enum AgentMessage {
    System {
        content: String,
    },
    User {
        content: String,
    },
    Assistant {
        content: AssistantContent,
    },
    Tool {
        tool_call_id: String,
        content: String,
    },
}

A single message in the agent conversation.

The closed enum prevents invalid states that a flat struct with role: String would allow (e.g. a User message carrying tool_calls, or a Tool message without a tool_call_id).

Wire format

#[serde(tag = "role", rename_all = "lowercase")] produces JSON identical to the TypeScript ChatMessage interface in the frontend:

{ "role": "user", "content": "What files are in the project?" }
{ "role": "assistant", "content": null, "tool_calls": [...] }
{ "role": "tool", "tool_call_id": "call_abc", "content": "src/\nlib/" }

Variants

System

A system-level instruction that sets the model’s persona and constraints.

Fields

content: String

Instruction text.

User

A message from the human user.

Fields

content: String

Message text.

Assistant

A response from the assistant model.

content always carries either text, tool calls, or both — the vacuous all-None state of the previous Option<String> + Option<Vec<ToolCall>> representation is impossible to construct.

Fields

content: AssistantContent

Content of the assistant turn.

Tool

The result of a tool call, to be sent back to the model.

Fields

tool_call_id: String

Must match the ToolCall::id from the preceding Assistant message.

content: String

Serialised output of the tool (or error description if it failed).

Implementations

impl AgentMessage

pub fn char_count(&self) -> usize

Estimate the Unicode scalar-value count of this message.

Uses [str::chars().count()] rather than str::len (byte count) so that multi-byte characters are counted as one unit, matching how LLMs typically measure context length.

Performance

This is an O(n) scan — it iterates over every Unicode scalar value in every str field of the message. Avoid calling it inside tight or nested loops. For repeated measurements over the same message set, accumulate the total once and update it incrementally (the agent loop does exactly this via its running_chars counter).

Trait Implementations

impl Clone for AgentMessage

fn clone(&self) -> AgentMessage

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for AgentMessage

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for AgentMessage

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for AgentMessage

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

where __S: Serializer,

Serialize this value into the given Serde serializer.