Pi uses AgentMessage values in SDK state, lifecycle events, RPC responses, and persisted session message entries. This page defines those shared messages and their content blocks.
Message timestamps are Unix timestamps in milliseconds. They are different from the ISO 8601 timestamps on session entries.
Source definitions:
packages/ai/src/types.tsdefines provider-facing messages and content blocks.packages/agent/src/types.tsdefines the extensibleAgentMessageunion.packages/coding-agent/src/core/messages.tsadds coding-agent message roles.
interface TextContent {
type: "text";
text: string;
textSignature?: string;
}textSignature contains provider-specific message metadata. Treat it as opaque.
interface ImageContent {
type: "image";
data: string;
mimeType: string;
}data is base64-encoded image data. mimeType identifies its media type, such as image/png or image/jpeg.
interface ThinkingContent {
type: "thinking";
thinking: string;
thinkingSignature?: string;
redacted?: boolean;
}Thinking signatures contain provider-specific replay data. Treat them as opaque. A redacted block can have no visible thinking text while retaining an encrypted payload in thinkingSignature.
interface ToolCall {
type: "toolCall";
id: string;
name: string;
arguments: Record<string, any>;
thoughtSignature?: string;
namespace?: string;
}thoughtSignature is provider-specific. namespace identifies an OpenAI Responses namespace for dynamically loaded or namespaced tools.
Assistant messages always contain usage. Tool results can contain usage when the tool performed nested model work.
interface Usage {
input: number;
output: number;
cacheRead: number;
cacheWrite: number;
cacheWrite1h?: number;
reasoning?: number;
totalTokens: number;
cost: {
input: number;
output: number;
cacheRead: number;
cacheWrite: number;
total: number;
};
}When present, reasoning is already included in output; do not add it again. cacheWrite1h is the subset of cacheWrite written with one-hour retention.
interface SystemMessage {
role: "system";
content: string | TextContent[];
sections?: Record<string, string | null>;
toolsAdded?: Tool[];
toolsRemoved?: ToolReference[];
replace?: boolean;
timestamp: number;
}The leading system message declares the initial prompt and tools. Later system messages can append instructions, replace or remove named prompt sections, and add or remove tools. Replaying them in order yields the current state. A message with replace: true discards the earlier state and establishes a complete new baseline.
interface UserMessage {
role: "user";
content: string | (TextContent | ImageContent)[];
timestamp: number;
}interface AssistantMessage {
role: "assistant";
content: (TextContent | ThinkingContent | ToolCall)[];
api: string;
provider: string;
model: string;
responseModel?: string;
responseId?: string;
providerThinkingLevel?: string;
diagnostics?: AssistantMessageDiagnostic[];
usage: Usage;
stopReason: "pending" | "stop" | "length" | "toolUse" | "error" | "aborted" | "deferred";
deferred?: DeferredHandle;
errorMessage?: string;
rawStopReason?: string;
endTurn?: boolean;
timestamp: number;
}responseModel records a concrete provider response model when it differs from the requested model. responseId, providerThinkingLevel, diagnostics, and rawStopReason preserve provider or runtime details.
"pending" is used for a partial assistant message while it streams. The completed message in message_end has a terminal stop reason, and Pi does not persist "pending" assistant messages in session JSONL.
A "deferred" response has a DeferredHandle with the provider data needed to retrieve it:
interface DeferredHandle {
provider: string;
modelId: string;
api: string;
id: string;
expiresAt?: number;
pollAfterMs?: number;
data?: JsonValue;
}interface ToolResultMessage<TDetails = any> {
role: "toolResult";
toolCallId: string;
toolName: string;
content: (TextContent | ImageContent)[];
details?: TDetails;
usage?: Usage;
isError: boolean;
timestamp: number;
}details is tool-specific. Optional usage reports nested model work performed by the tool and contributes to full-session statistics, but it is not part of the main model-call usage.
The coding-agent package extends AgentMessage with four roles.
Created by direct shell commands, including the RPC bash command. It is not an LLM tool result.
interface BashExecutionMessage {
role: "bashExecution";
command: string;
output: string;
exitCode: number | undefined;
cancelled: boolean;
truncated: boolean;
fullOutputPath?: string;
excludeFromContext?: boolean;
timestamp: number;
}Unless excludeFromContext is true, Pi converts this message to user-role text before the next model request.
Created when an extension sends a context message.
interface CustomMessage<T = unknown> {
role: "custom";
customType: string;
content: string | (TextContent | ImageContent)[];
display: boolean;
details?: T;
timestamp: number;
}Pi converts its content to a user message for model requests. display controls terminal rendering; details is not sent to the model.
interface BranchSummaryMessage {
role: "branchSummary";
summary: string;
fromId: string | null;
timestamp: number;
}Pi creates this context message from a persisted branch_summary entry.
interface CompactionSummaryMessage {
role: "compactionSummary";
summary: string;
tokensBefore: number;
timestamp: number;
}Pi creates this context message from a persisted compaction entry.
In the coding agent, the union is equivalent to:
type AgentMessage =
| SystemMessage
| UserMessage
| AssistantMessage
| ToolResultMessage
| BashExecutionMessage
| CustomMessage
| BranchSummaryMessage
| CompactionSummaryMessage;At the lower-level agent package, AgentMessage is Message | CustomAgentMessages[keyof CustomAgentMessages]. Applications can add roles through TypeScript declaration merging, so consumers should tolerate unknown custom roles when they accept messages from an augmented host.