Documentation
¶
Overview ¶
Package api provides the HTTP client for communicating with the Agentic Serving Layer daemon (RFC 5).
Package api provides the HTTP client for communicating with the Agentic Serving Layer daemon (RFC 5).
Package api provides the HTTP API for the Agentic Serving Layer daemon (RFC 5). The daemon is a control plane for coordination (shared context, cache policies, workflow execution). Actual inference happens on the LLM backends; the daemon coordinates state across multiple agents and LLM backends.
Index ¶
- type AgentExecuteRequest
- type AgentExecuteResponse
- type Client
- type CompleteTaskRequest
- type CompleteTaskResponse
- type ContextSyncRequest
- type ContextSyncResponse
- type DaemonCoordinator
- func (c *DaemonCoordinator) CompleteTask(ctx context.Context, executionID string, taskIndex int, ...) error
- func (c *DaemonCoordinator) GetContext(ctx context.Context, serverName string) ([]model.Message, error)
- func (c *DaemonCoordinator) GetExecution(ctx context.Context, executionID string) (*serving.WorkflowExecution, error)
- func (c *DaemonCoordinator) GetNextTask(ctx context.Context, executionID string) (*config.WorkflowTask, int, bool, error)
- func (c *DaemonCoordinator) Health(ctx context.Context) error
- func (c *DaemonCoordinator) StartWorkflow(ctx context.Context, workflowName string) (string, error)
- func (c *DaemonCoordinator) SyncContext(ctx context.Context, serverName string, messages []model.Message) error
- type ExecuteTaskRequest
- type ExecuteTaskRequestOptions
- type ExecuteTaskResponse
- type GetContextResponse
- type GetExecutionResponse
- type GetNextTaskResponse
- type Server
- type WorkflowStartRequest
- type WorkflowStartResponse
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type AgentExecuteRequest ¶
type AgentExecuteRequest struct {
ProfileName string `json:"profile_name"`
Prompt string `json:"prompt"`
Messages []model.Message `json:"messages,omitempty"` // Conversation history
Tools []*mcp.Tool `json:"tools,omitempty"` // Available tools (MCP format)
MaxTokens *int `json:"max_tokens,omitempty"`
Stream bool `json:"stream,omitempty"`
}
AgentExecuteRequest represents a request to execute an agent
type AgentExecuteResponse ¶
type AgentExecuteResponse struct {
Success bool `json:"success"`
Response *model.Response `json:"response,omitempty"`
Error string `json:"error,omitempty"`
}
AgentExecuteResponse represents the response from agent execution
type Client ¶
type Client struct {
// contains filtered or unexported fields
}
Client is the HTTP client for communicating with the daemon
func (*Client) StartWorkflow ¶
func (c *Client) StartWorkflow(ctx context.Context, workflowName string) (*WorkflowStartResponse, error)
StartWorkflow starts a workflow execution
type CompleteTaskRequest ¶
type CompleteTaskRequest struct {
ExecutionID string `json:"execution_id"`
TaskIndex int `json:"task_index"`
Response *model.Response `json:"response,omitempty"`
}
CompleteTaskRequest represents a request to complete a task
type CompleteTaskResponse ¶
type CompleteTaskResponse struct {
Success bool `json:"success"`
Error string `json:"error,omitempty"`
}
CompleteTaskResponse represents the response from completing a task
type ContextSyncRequest ¶
type ContextSyncRequest struct {
ServerName string `json:"server_name"`
Messages []model.Message `json:"messages"`
}
ContextSyncRequest represents a request to sync context
type ContextSyncResponse ¶
type ContextSyncResponse struct {
Success bool `json:"success"`
Error string `json:"error,omitempty"`
}
ContextSyncResponse represents the response from context sync
type DaemonCoordinator ¶
type DaemonCoordinator struct {
// contains filtered or unexported fields
}
DaemonCoordinator handles communication with the daemon for coordination (shared context, workflow management). It does NOT proxy inference requests. Actual inference happens locally via providers; the daemon coordinates state.
func NewDaemonCoordinator ¶
func NewDaemonCoordinator(daemonURL string) *DaemonCoordinator
NewDaemonCoordinator creates a new daemon coordinator
func (*DaemonCoordinator) CompleteTask ¶
func (c *DaemonCoordinator) CompleteTask(ctx context.Context, executionID string, taskIndex int, response *model.Response) error
CompleteTask marks a task as complete and reports the response
func (*DaemonCoordinator) GetContext ¶
func (c *DaemonCoordinator) GetContext(ctx context.Context, serverName string) ([]model.Message, error)
GetContext retrieves shared context from the daemon
func (*DaemonCoordinator) GetExecution ¶
func (c *DaemonCoordinator) GetExecution(ctx context.Context, executionID string) (*serving.WorkflowExecution, error)
GetExecution retrieves workflow execution details from the daemon
func (*DaemonCoordinator) GetNextTask ¶
func (c *DaemonCoordinator) GetNextTask(ctx context.Context, executionID string) (*config.WorkflowTask, int, bool, error)
GetNextTask retrieves the next task to execute from a workflow
func (*DaemonCoordinator) Health ¶
func (c *DaemonCoordinator) Health(ctx context.Context) error
Health checks daemon health
func (*DaemonCoordinator) StartWorkflow ¶
StartWorkflow starts a workflow execution on the daemon
func (*DaemonCoordinator) SyncContext ¶
func (c *DaemonCoordinator) SyncContext(ctx context.Context, serverName string, messages []model.Message) error
SyncContext syncs the local context with the daemon This is called after a task execution to share context updates
type ExecuteTaskRequest ¶
type ExecuteTaskRequest struct {
ExecutionID string `json:"execution_id"`
TaskIndex int `json:"task_index"`
Prompt string `json:"prompt"`
Options *ExecuteTaskRequestOptions `json:"options,omitempty"` // Optional; mirrors serving.ExecuteTaskOptions
}
ExecuteTaskRequest represents a request to execute a task
type ExecuteTaskRequestOptions ¶ added in v1.1.0
type ExecuteTaskRequestOptions struct {
MaxTokens *int `json:"max_tokens,omitempty"` // Optional: maximum tokens to generate
Stream bool `json:"stream,omitempty"` // If true, use streaming and populate response.metrics (ttft_ms, tpot_ms)
}
ExecuteTaskRequestOptions are optional parameters for task execution (mirrors serving.ExecuteTaskOptions).
type ExecuteTaskResponse ¶
type ExecuteTaskResponse struct {
Success bool `json:"success"`
Response *model.Response `json:"response,omitempty"`
Error string `json:"error,omitempty"`
}
ExecuteTaskResponse represents the response from executing a task
type GetContextResponse ¶
type GetContextResponse struct {
Messages []model.Message `json:"messages"`
Error string `json:"error,omitempty"`
}
GetContextResponse represents the response from get context
type GetExecutionResponse ¶
type GetExecutionResponse struct {
Execution *serving.WorkflowExecution `json:"execution,omitempty"`
Error string `json:"error,omitempty"`
}
GetExecutionResponse represents the response from get execution
type GetNextTaskResponse ¶
type GetNextTaskResponse struct {
Task *config.WorkflowTask `json:"task,omitempty"`
TaskIndex int `json:"task_index"`
Complete bool `json:"complete"`
LLMServer string `json:"llm_server,omitempty"` // Resolved server name (from profile or task override)
Error string `json:"error,omitempty"`
}
GetNextTaskResponse represents the response from get next task
type Server ¶
type Server struct {
// contains filtered or unexported fields
}
Server is the HTTP API server for the daemon
func NewServer ¶
func NewServer(servingLayer serving.ServingLayer, listenAddress string) *Server
NewServer creates a new daemon API server
type WorkflowStartRequest ¶
type WorkflowStartRequest struct {
WorkflowName string `json:"workflow_name"`
}
WorkflowStartRequest represents a workflow start request
type WorkflowStartResponse ¶
type WorkflowStartResponse struct {
ExecutionID string `json:"execution_id"`
Error string `json:"error,omitempty"`
}
WorkflowStartResponse represents a workflow start response
Source Files