agent

package
v6.13.0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Aug 30, 2026 License: Apache-2.0 Imports: 37 Imported by: 0

Documentation

Overview

Package agent provides the Agent abstraction for Go Micro.

An Agent is a service with an LLM inside it. It registers a Chat RPC endpoint, discovers its assigned services' tools, and orchestrates them intelligently. 

agent := micro.NewAgent("task-mgr",
    micro.AgentServices("task"),
    micro.AgentPrompt("You manage tasks."),
    micro.AgentProvider("anthropic"),
)
agent.Run()

Index

Constants

View Source 
const (
	AttrRunID            = "agent.run.id"
	AttrParentRunID      = "agent.run.parent_id"
	AttrAgentName        = "agent.name"
	AttrProvider         = "agent.model.provider"
	AttrModel            = "agent.model.name"
	AttrLatencyMS        = "agent.latency_ms"
	AttrInputTokens      = "agent.tokens.input"
	AttrOutputTokens     = "agent.tokens.output"
	AttrTotalTokens      = "agent.tokens.total"
	AttrAttempt          = "agent.model.attempt"
	AttrMaxAttempts      = "agent.model.max_attempts"
	AttrToolAttempt      = "agent.tool.attempt"
	AttrToolMaxAttempts  = "agent.tool.max_attempts"
	AttrToolName         = "agent.tool.name"
	AttrDelegate         = "agent.delegate"
	AttrGuardrailBlock   = "agent.guardrail.block"
	AttrRefusal          = "agent.refusal"
	AttrInputChars       = "agent.input.chars"
	AttrErrorKind        = "agent.error.kind"
	AttrCheckpointStatus = "agent.checkpoint.status"
	AttrCheckpointStage  = "agent.checkpoint.stage"
	AttrFlowName         = "agent.flow.name"
	AttrFlowStep         = "agent.flow.step"
	AttrDispatch         = "agent.dispatch"
	AttrTrigger          = "agent.trigger"
	AttrRunEventKind     = "agent.event.kind"
	AttrSpend            = "agent.spend"
	AttrToolSpend        = "agent.tool.spend"
)

Variables

This section is empty.

Functions

func Builtins 

func Builtins(opts ...Option) (tools []ai.Tool, handle func(name string, input map[string]any) (result any, content string, ok bool))

Builtins returns the built-in agent tools (plan, delegate) together with a handler for them, so the same capabilities can be wired into a tool loop that isn't a running Agent — for example the `micro chat` fallback. The handler's third return value is false when the name is not a built-in, so callers can fall through to their own tools.

Configure it with the same options as an Agent (Name, Provider, WithStore, WithRegistry, WithClient, ...); these back plan's memory and delegate's RPC/sub-agent behavior.

func Pending added in v6.3.1 

func Pending(ctx context.Context, ag Agent) ([]flow.Run, error)

Pending returns checkpointed agent runs that have not completed. It mirrors flow.Pending for startup recovery loops that drain durable agent work.

func ResumePending added in v6.3.18 

func ResumePending(ctx context.Context, ag Agent) (string, error)

ResumePending resumes every checkpointed agent run that has not completed yet, in the same oldest-first order returned by Pending.

It is a convenience for service startup and recovery loops: after recreating an agent with the same checkpoint store, call ResumePending to drain the durable backlog without listing and resuming each run manually. If any run fails again, ResumePending stops and returns that run id with the error so callers can log, alert, or retry later without hiding the failing run.

func Summary added in v6.7.1 

func Summary(m Memory) string

Summary returns the current compacted-memory summary for m, when supported. It returns an empty string for memory backends that have not compacted or do not expose an inspectable summary.

Types

type Agent 

type Agent interface {
	Name() string
	Init(...Option)
	Options() Options
	Ask(ctx context.Context, message string) (*Response, error)
	Stream(ctx context.Context, message string) (ai.Stream, error)
	Run() error
	Stop() error
	String() string
}

Agent is the interface for an AI agent that manages services.

func New 

func New(opts ...Option) Agent

New creates a new Agent.

type AgentStream added in v6.3.9 

type AgentStream interface {
	Recv() (*StreamEvent, error)
	Close() error
}

AgentStream is a stream of tool execution events followed by final-answer chunks.

func ResumeStreamAsk added in v6.3.9 

func ResumeStreamAsk(ctx context.Context, ag Agent, runID string) (AgentStream, error)

ResumeStreamAsk resumes a checkpointed agent run and emits the same event shape as StreamAsk. Completed runs are streamed from the persisted response; unfinished runs continue from their checkpoint and emit tool events for any work that still needs to run. Tool calls already recorded as done in the checkpoint are reused by the agent checkpoint wrapper and are not re-executed.

func StreamAsk added in v6.3.9 

func StreamAsk(ctx context.Context, ag Agent, message string) (AgentStream, error)

StreamAsk runs an agent Ask turn with tool start/end events and streams the final answer. It is additive for callers that hold the public Agent interface; concrete agents also expose the same method directly.

type ApproveFunc 

type ApproveFunc func(tool string, input map[string]any) (approved bool, reason string)

ApproveFunc decides whether an agent may execute a tool call before it runs. Returning false blocks the call; the reason is shown to the model so it can adapt. Use it for human-in-the-loop approval or policy checks. It is called for actions (service tools and delegate), not for the internal plan tool.

type Memory 

type Memory interface {
	// Add appends a message to the conversation.
	Add(role, content string)
	// Messages returns the retained conversation, oldest first.
	Messages() []ai.Message
	// Clear resets the conversation.
	Clear()
}

Memory is an agent's conversation memory. Like the rest of the framework it is pluggable: the default is store-backed and durable across restarts, but any implementation can be supplied with WithMemory — in-process, a database, or a semantic/vector store.

func NewCompactingMemory added in v6.3.1 

func NewCompactingMemory(s store.Store, key string, maxMessages, keepRecent int) Memory

NewCompactingMemory returns store-backed memory with explicit compaction and retrieval controls. It keeps all messages in the backing store, compacts older turns into a deterministic summary when the conversation exceeds maxMessages, and lets callers recall relevant prior turns with Recall.

func NewCompactingMemoryWithOptions added in v6.3.11 

func NewCompactingMemoryWithOptions(s store.Store, key string, compaction MemoryCompaction) Memory

NewCompactingMemoryWithOptions returns store-backed memory configured with explicit compaction options, including an optional summarization hook.

func NewInMemory 

func NewInMemory(limit int) Memory

NewInMemory returns conversation memory that is not persisted.

func NewMemory 

func NewMemory(s store.Store, key string, limit int) Memory

NewMemory returns the default store-backed memory: an in-process conversation buffer (truncated to limit) that persists to the store under key, so an agent picks up where it left off after a restart. A nil store or empty key yields non-persistent memory.

func NewRetrievalMemory added in v6.3.11 

func NewRetrievalMemory(s store.Store, key string, activeLimit int) Memory

NewRetrievalMemory returns store-backed memory that keeps a bounded active conversation and archives every turn for retrieval. It is useful when callers want relevant durable recall without summary compaction in the active context. A nil store or empty key keeps only the active in-process buffer.

type MemoryCompaction added in v6.3.1 

type MemoryCompaction struct {
	MaxMessages int
	KeepRecent  int
	Summarize   MemorySummaryFunc
}

MemoryCompaction configures deterministic, store-backed context compaction for the default memory implementation. When the retained conversation grows past MaxMessages, older turns are collapsed into a summary message while the newest KeepRecent turns stay verbatim for provider-neutral continuity.

type MemoryRecall added in v6.3.1 

type MemoryRecall interface {
	Recall(query string, limit int) []ai.Message
}

MemoryRecall is implemented by memory backends that can retrieve durable prior context relevant to a new turn without replaying every stored message.

type MemorySummary added in v6.7.1 

type MemorySummary interface {
	Summary() string
}

MemorySummary is implemented by memory backends that expose their current compacted summary for inspection. It lets long-running agents make memory compaction observable without coupling callers to a concrete store.

type MemorySummaryFunc added in v6.3.11 

type MemorySummaryFunc func([]ai.Message) ai.Message

MemorySummaryFunc turns older conversation messages into a compact replacement message for active context. It is called while the default memory is locked, so implementations should be deterministic and avoid calling back into the same memory instance.

type Option 

type Option func(*Options)

Option configures an Agent.

func APIKey 

func APIKey(k string) Option

APIKey sets the API key for the LLM provider.

func Address added in v6.3.1 

func Address(addr string) Option

Address sets the network address for the agent's service endpoint. Use "127.0.0.1:0" in local harnesses/tests to bind an ephemeral loopback port and avoid advertising the default service address.

func ApproveTool 

func ApproveTool(fn ApproveFunc) Option

ApproveTool sets a human-in-the-loop / policy hook called before each action (service tools and delegate). Returning false blocks the call.

func BaseURL added in v6.3.12 

func BaseURL(url string) Option

BaseURL sets the base URL for the LLM provider. Use this to point the provider at a non-default endpoint (e.g., local Ollama, a proxy).

func Budget added in v6.7.1 

func Budget(amount int64) Option

Budget bounds autonomous x402 payments per Ask, in the asset's smallest unit (0 = unlimited). The budget is enforced by wrapper/x402.Client.

func CompactMemory added in v6.3.1 

func CompactMemory(maxMessages, keepRecent int) Option

CompactMemory enables deterministic, store-backed memory compaction for the default agent memory. Older turns are summarized once active context exceeds maxMessages, keepRecent newest turns remain verbatim, and recalled archived turns are injected into matching future asks.

func HistoryLimit 

func HistoryLimit(n int) Option

HistoryLimit sets the max conversation messages to retain.

func LoopLimit 

func LoopLimit(n int) Option

LoopLimit sets how many times the agent may repeat the same tool call (same name and arguments) in one Ask before it is refused as a no-progress loop. 0 disables loop detection.

func MaxSpend added in v6.7.0 

func MaxSpend(amount int64) Option

MaxSpend bounds paid x402 tool spend per Ask, in the asset's smallest unit (0 = disabled). A paid tool that would exceed the cap is refused before the tool handler runs or any payment can be made.

func MaxSteps 

func MaxSteps(n int) Option

MaxSteps bounds tool executions per Ask (0 = unbounded). A stopping condition: beyond the limit, tool calls are refused and the model is told to stop and summarize.

func MemoryRecallLimit added in v6.3.1 

func MemoryRecallLimit(n int) Option

MemoryRecallLimit sets how many archived turns a memory backend may inject into a model request for the current Ask. Use 0 to disable retrieval.

func MemorySummarizer added in v6.3.11 

func MemorySummarizer(fn MemorySummaryFunc) Option

MemorySummarizer sets the deterministic summarization hook used by the default compacting memory. It is optional; without it, compacted memory uses a provider-neutral text summary. The hook receives the older messages being removed from active context and returns the replacement summary message.

func Model 

func Model(m string) Option

Model sets the LLM model name.

func ModelCallTimeout added in v6.3.0 

func ModelCallTimeout(d time.Duration) Option

ModelCallTimeout sets the timeout for each provider Generate call.

func ModelRetry added in v6.3.0 

func ModelRetry(maxAttempts int, backoff time.Duration) Option

ModelRetry sets the provider retry budget and backoff for transient failures.

func ModelRetryJitter added in v6.7.1 

func ModelRetryJitter(d time.Duration) Option

ModelRetryJitter adds bounded random jitter to provider retry backoff. Set 0 to disable.

func Name 

func Name(n string) Option

Name sets the agent name.

func Payer added in v6.7.1 

func Payer(p x402.Payer) Option

Payer configures the wallet/signing hook used to settle x402-paid tools. Without a payer, payment-required tool results are returned as clear errors.

func Prompt 

func Prompt(p string) Option

Prompt sets the system prompt.

func Provider 

func Provider(p string) Option

Provider sets the LLM provider.

func RetrievalMemory added in v6.3.11 

func RetrievalMemory(activeLimit int) Option

RetrievalMemory enables deterministic, store-backed retrieval memory for the default agent memory without compaction. Active context is capped at activeLimit messages while every turn is archived in the store for Recall.

func Services 

func Services(names ...string) Option

Services sets which services this agent manages.

func ToolCallTimeout added in v6.3.11 

func ToolCallTimeout(d time.Duration) Option

ToolCallTimeout sets the timeout for each tool execution. It bounds custom tools, built-in delegate calls, and service RPC tools with the same context deadline so mid-run cancellation and slow tools produce safe error results instead of unbounded agent runs. Set 0 to disable.

func ToolRetry added in v6.3.11 

func ToolRetry(maxAttempts int, backoff time.Duration) Option

ToolRetry sets the tool retry budget and backoff for transient failures. Attempts include the first call. Retries are opt-in because tools may have side effects; keep handlers idempotent before enabling this.

func ToolSpend added in v6.7.0 

func ToolSpend(tool string, amount int64) Option

ToolSpend records the x402 price for a tool, in the asset's smallest unit, so MaxSpend can reserve budget before execution. Non-positive amounts are treated as free.

func TraceInputs added in v6.3.11 

func TraceInputs(enabled bool) Option

TraceInputs opts in to recording raw user messages on agent run events. By default inputs are redacted from OpenTelemetry spans and persisted run timelines; use this only when the observability backend is approved to store prompt content.

func TraceProvider added in v6.3.0 

func TraceProvider(tp trace.TracerProvider) Option

TraceProvider enables OpenTelemetry tracing for agent runs. The persisted run timeline is recorded even when TraceProvider is nil; trace/span IDs are added only when a provider is configured.

func WithA2A 

func WithA2A(addr string) Option

WithA2A makes Run serve the agent over the A2A protocol on addr (e.g. ":4000"), so other agents can reach it directly by URL without a separate gateway. The agent stays a normal go-micro service as well; this adds a second, A2A-native HTTP endpoint that calls it in-process.

func WithBroker added in v6.3.15 

func WithBroker(b broker.Broker) Option

WithBroker sets the broker used by the agent service endpoint. Use an in-memory broker in local harnesses/tests to avoid sharing the package-wide default broker listener across concurrently running examples.

func WithCheckpoint added in v6.3.1 

func WithCheckpoint(c flow.Checkpoint) Option

WithCheckpoint sets the durability backend for agent Ask runs. The Checkpoint interface is shared with flow so services, agents, and workflows can use one execution history backend. When set, each Ask is saved as a single-step run keyed by run id; Resume returns a completed run's persisted response instead of calling the model again.

func WithClient 

func WithClient(c client.Client) Option

WithClient sets the RPC client.

func WithMemory 

func WithMemory(m Memory) Option

WithMemory sets the agent's conversation memory. The default is store-backed memory keyed by agent name; supply your own to use an in-process, database, or semantic store.

func WithRegistry 

func WithRegistry(r registry.Registry) Option

WithRegistry sets the service registry.

func WithStore 

func WithStore(s store.Store) Option

WithStore sets the store for agent memory.

func WithTool 

func WithTool(name, description string, properties map[string]any, handler ToolFunc) Option

WithTool registers a custom tool the agent can call, beyond the services it discovers — a local function, an external API, anything. properties is the JSON-schema map for the tool's parameters.

func WrapTool 

func WrapTool(w ...ai.ToolWrapper) Option

WrapTool registers a tool-execution wrapper, the tool-side analog of a client/server middleware wrapper. Each wrapper takes the next handler and returns a new one; code before the next(...) call runs before the tool executes, code after runs after. Use it for logging, metrics, retries, or custom policy. Wrappers run outside the built-in guardrails (MaxSteps, LoopLimit, ApproveTool), so they observe every call and its result, including refusals. Multiple wrappers compose outermost-first. 

micro.NewAgent("worker", micro.AgentWrapTool(
    func(next ai.ToolHandler) ai.ToolHandler {
        return func(ctx context.Context, call ai.ToolCall) ai.ToolResult {
            res := next(ctx, call)
            log.Printf("id=%s tool=%s", call.ID, call.Name)
            return res
        }
    }))

type Options 

type Options struct {
	Name         string
	Services     []string
	Prompt       string
	Provider     string
	Model        string
	APIKey       string
	BaseURL      string
	Address      string
	Registry     registry.Registry
	Client       client.Client
	Broker       broker.Broker
	Store        store.Store
	HistoryLimit int

	// ModelTimeout bounds each provider Generate call (0 disables).
	ModelTimeout time.Duration
	// ModelMaxAttempts bounds provider Generate attempts including the first
	// call. Default 1 — retries are opt-in (enable with ModelRetry). A Generate
	// runs the whole tool-execution turn, so auto-retrying it would re-run
	// already-executed, possibly side-effecting tool calls; keep it explicit.
	ModelMaxAttempts int
	// ModelRetryBackoff is the base delay between transient provider failures
	// (grows exponentially per attempt when retries are enabled).
	ModelRetryBackoff time.Duration
	// ModelRetryJitter adds up to this random delay to each provider retry
	// backoff. Default 0 preserves deterministic timing unless explicitly set.
	ModelRetryJitter time.Duration
	// ToolTimeout bounds each tool execution (0 disables). The timeout is
	// applied before custom tools, delegate, and service RPC calls so context
	// deadlines propagate consistently through the agent loop.
	ToolTimeout time.Duration
	// ToolMaxAttempts bounds tool execution attempts including the first call.
	// Default 1; retries are opt-in because tools can have side effects.
	ToolMaxAttempts int
	// ToolRetryBackoff is the base delay between transient tool failures.
	ToolRetryBackoff time.Duration

	// Memory is the agent's conversation memory. Nil = the default
	// store-backed memory (durable across restarts).
	Memory Memory
	// MemoryRetrievalLimit enables retrieval-backed default memory without
	// compaction. The active conversation stays bounded to this many messages
	// while every turn is archived for deterministic recall.
	MemoryRetrievalLimit int
	// MemoryCompaction enables deterministic compaction/retrieval on the
	// default store-backed memory. Custom Memory implementations can expose
	// retrieval by implementing MemoryRecall.
	MemoryCompaction MemoryCompaction
	// MemoryRecallLimit bounds recalled archived turns injected into a model
	// request (0 disables recall injection).
	MemoryRecallLimit int
	// Checkpoint persists agent Ask runs so callers can resume by run id
	// after a restart without replaying a run that already completed.
	Checkpoint flow.Checkpoint

	// MaxSteps bounds the number of tool executions per Ask (0 =
	// unbounded). Once exceeded, further tool calls are refused and the
	// model is told to stop and summarize. A stopping condition.
	MaxSteps int
	// LoopLimit bounds how many times the agent may call the same tool
	// with the same arguments in one Ask before the call is refused as a
	// no-progress loop (0 = disabled). Catches the agent repeating an
	// identical action — which MaxSteps only bounds by total count.
	LoopLimit int
	// Approve gates each action before it runs. Nil = allow all.
	Approve ApproveFunc
	// MaxSpend bounds paid x402 tool spend per Ask in the asset's smallest
	// unit (0 = disabled). ToolSpend lists known paid tools and their prices.
	MaxSpend  int64
	ToolSpend map[string]int64
	// Payer lets the agent settle x402 Payment Required challenges from tools.
	// Budget bounds autonomous x402 payments per Ask (0 = unlimited).
	Payer  x402.Payer
	Budget int64

	// A2AAddress, if set, makes Run serve this agent over the A2A protocol
	// on that address directly (no separate gateway), e.g. ":4000".
	A2AAddress string

	// TraceProvider enables OpenTelemetry spans for agent runs, model calls,
	// and tool calls. Nil disables instrumentation.
	TraceProvider trace.TracerProvider

	// TraceInputs controls whether agent observability records include raw
	// user messages. It is false by default so spans and persisted run
	// timelines carry correlation and shape without leaking prompts.
	TraceInputs bool
	// contains filtered or unexported fields
}

Options holds agent configuration.

type Response 

type Response struct {
	Reply     string
	ToolCalls []ai.ToolCall
	Agent     string

	// RunID correlates this Ask with tool calls, trace spans, and the
	// persisted run timeline. ParentID is set when this response belongs
	// to a delegated sub-agent run.
	RunID    string
	ParentID string
}

Response is what an agent returns from Chat.

func Resume added in v6.3.1 

func Resume(ctx context.Context, ag Agent, runID string) (*Response, error)

Resume returns the response for a checkpointed agent run. Completed runs are returned from the checkpoint without calling the model or replaying tool calls; failed or in-progress runs continue from the saved input message.

func ResumeInput added in v6.3.1 

func ResumeInput(ctx context.Context, ag Agent, runID, input string) (*Response, error)

ResumeInput resumes a checkpointed agent run that paused via the built-in request_input tool. The supplied input is appended to the original request so the same run can continue with durable checkpoint and completed tool history.

type RunEvent added in v6.3.0 

type RunEvent struct {
	Time        time.Time `json:"time"`
	RunID       string    `json:"run_id"`
	ParentID    string    `json:"parent_id,omitempty"`
	TraceID     string    `json:"trace_id,omitempty"`
	SpanID      string    `json:"span_id,omitempty"`
	Agent       string    `json:"agent"`
	Kind        string    `json:"kind"`
	Name        string    `json:"name,omitempty"`
	Provider    string    `json:"provider,omitempty"`
	Model       string    `json:"model,omitempty"`
	Attempt     int       `json:"attempt,omitempty"`
	MaxAttempts int       `json:"max_attempts,omitempty"`
	LatencyMS   int64     `json:"latency_ms,omitempty"`
	Tokens      Usage     `json:"tokens,omitempty"`
	Refused     string    `json:"refused,omitempty"`
	Status      string    `json:"status,omitempty"`
	Error       string    `json:"error,omitempty"`
	ErrorKind   string    `json:"error_kind,omitempty"`
	InputChars  int       `json:"input_chars,omitempty"`
	Spent       int64     `json:"spent,omitempty"`
	ToolSpend   int64     `json:"tool_spend,omitempty"`
}

func LoadRunEvents added in v6.3.0 

func LoadRunEvents(s store.Store, agentName, runID string) ([]RunEvent, error)

type RunListOptions added in v6.3.1 

type RunListOptions struct {
	// Status, when set, keeps only runs with the matching status
	// (for example "running", "done", "canceled", "timeout",
	// "rate_limited", "auth", "configuration", "unavailable",
	// "provider_error", "error", or "refused").
	Status string
	// TraceID, when set, keeps only runs correlated with this trace id.
	// A prefix is accepted so operators can paste the shortened trace id
	// printed by `micro runs`.
	TraceID string
	// Limit, when positive, returns the most recently updated runs up to
	// the limit. Limited results are ordered newest first.
	Limit int
}

RunListOptions controls how recorded agent run summaries are returned. Zero values preserve the full deterministic run list.

type RunSummary added in v6.3.1 

type RunSummary struct {
	RunID         string    `json:"run_id"`
	Agent         string    `json:"agent"`
	ParentID      string    `json:"parent_id,omitempty"`
	TraceID       string    `json:"trace_id,omitempty"`
	SpanID        string    `json:"span_id,omitempty"`
	StartedAt     time.Time `json:"started_at"`
	UpdatedAt     time.Time `json:"updated_at"`
	DurationMS    int64     `json:"duration_ms,omitempty"`
	Events        int       `json:"events"`
	Status        string    `json:"status,omitempty"`
	Checkpoint    string    `json:"checkpoint,omitempty"`
	Stage         string    `json:"stage,omitempty"`
	LastKind      string    `json:"last_kind,omitempty"`
	LastError     string    `json:"last_error,omitempty"`
	LastErrorKind string    `json:"last_error_kind,omitempty"`
	Spent         int64     `json:"spent,omitempty"`
}

RunSummary is a compact index entry for a recorded agent run.

func ListRunSummaries added in v6.3.1 

func ListRunSummaries(s store.Store, agentName string) ([]RunSummary, error)

ListRunSummaries returns a deterministic summary of recorded runs for agentName.

func ListRunSummariesWithOptions added in v6.3.1 

func ListRunSummariesWithOptions(s store.Store, agentName string, opts RunListOptions) ([]RunSummary, error)

ListRunSummariesWithOptions returns summaries of recorded runs for agentName, optionally filtered by status and limited to the most recently updated runs.

type StreamEvent added in v6.3.9 

type StreamEvent struct {
	Type     StreamEventType
	Token    string
	ToolCall ai.ToolCall
	Result   ai.ToolResult
	Response *Response
}

StreamEvent is one event from StreamAsk.

type StreamEventType added in v6.3.9 

type StreamEventType string

StreamEventType identifies an event emitted by a tool-aware agent stream. 

const (
	// StreamEventToolStart is emitted immediately before a tool call runs.
	StreamEventToolStart StreamEventType = "tool_start"
	// StreamEventToolEnd is emitted after a tool call returns or is refused.
	StreamEventToolEnd StreamEventType = "tool_end"
	// StreamEventToken carries a chunk of the final answer.
	StreamEventToken StreamEventType = "token"
	// StreamEventDone carries the completed agent response.
	StreamEventDone StreamEventType = "done"
)

type ToolFunc 

type ToolFunc func(ctx context.Context, input map[string]any) (string, error)

ToolFunc handles a custom tool call. Return the result as a string (often JSON); return an error to report failure back to the model.

type Usage added in v6.3.0 

type Usage = ai.Usage

Source Files

View all Source files

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.