agents

package
v0.3.4 Latest Latest
Warning

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

 Go to latest Published: Feb 28, 2026 License: MIT Imports: 10 Imported by: 0

Documentation

Overview

Package agents provides interfaces and implementations for spawning AI agents. Unlike providers (which track usage), agents execute tasks autonomously.

claude.go implements the Agent interface for Claude Code CLI.

codex.go implements the Agent interface for OpenAI Codex CLI.

copilot.go implements the Agent interface for GitHub Copilot CLI.

Index

Constants

View Source 
const DefaultTimeout = 30 * time.Minute

DefaultTimeout is the default agent execution timeout (30 minutes).

Variables

This section is empty.

Functions

This section is empty.

Types

type Agent 

type Agent interface {
	// Name returns the agent identifier.
	Name() string

	// Execute runs a prompt and returns the output.
	Execute(ctx context.Context, opts ExecuteOptions) (*ExecuteResult, error)
}

Agent is the interface for AI agent execution.

type ClaudeAgent 

type ClaudeAgent struct {
	// contains filtered or unexported fields
}

ClaudeAgent spawns Claude Code CLI for task execution.

func NewClaudeAgent 

func NewClaudeAgent(opts ...ClaudeOption) *ClaudeAgent

NewClaudeAgent creates a Claude Code agent.

func (*ClaudeAgent) Available 

func (a *ClaudeAgent) Available() bool

Available checks if the claude binary is available in PATH.

func (*ClaudeAgent) Execute 

func (a *ClaudeAgent) Execute(ctx context.Context, opts ExecuteOptions) (*ExecuteResult, error)

Execute runs claude --print with the given prompt.

func (*ClaudeAgent) ExecuteWithFiles 

func (a *ClaudeAgent) ExecuteWithFiles(ctx context.Context, prompt string, files []string, workDir string) (*ExecuteResult, error)

ExecuteWithFiles runs claude with file context included.

func (*ClaudeAgent) Name 

func (a *ClaudeAgent) Name() string

Name returns "claude".

func (*ClaudeAgent) Version 

func (a *ClaudeAgent) Version() (string, error)

Version returns the claude CLI version.

type ClaudeOption 

type ClaudeOption func(*ClaudeAgent)

ClaudeOption configures a ClaudeAgent.

func WithBinaryPath 

func WithBinaryPath(path string) ClaudeOption

WithBinaryPath sets a custom path to the claude binary.

func WithDangerouslySkipPermissions 

func WithDangerouslySkipPermissions(enabled bool) ClaudeOption

WithDangerouslySkipPermissions sets whether to pass --dangerously-skip-permissions.

func WithDefaultTimeout 

func WithDefaultTimeout(d time.Duration) ClaudeOption

WithDefaultTimeout sets the default execution timeout.

func WithRunner 

func WithRunner(r CommandRunner) ClaudeOption

WithRunner sets a custom command runner (for testing).

type CodexAgent 

type CodexAgent struct {
	// contains filtered or unexported fields
}

CodexAgent spawns Codex CLI for task execution.

func NewCodexAgent 

func NewCodexAgent(opts ...CodexOption) *CodexAgent

NewCodexAgent creates a Codex CLI agent.

func (*CodexAgent) Available 

func (a *CodexAgent) Available() bool

Available checks if the codex binary is available in PATH.

func (*CodexAgent) Execute 

func (a *CodexAgent) Execute(ctx context.Context, opts ExecuteOptions) (*ExecuteResult, error)

Execute runs codex with the given prompt in non-interactive mode.

func (*CodexAgent) ExecuteWithFiles 

func (a *CodexAgent) ExecuteWithFiles(ctx context.Context, prompt string, files []string, workDir string) (*ExecuteResult, error)

ExecuteWithFiles runs codex with file context included.

func (*CodexAgent) Name 

func (a *CodexAgent) Name() string

Name returns "codex".

func (*CodexAgent) Version 

func (a *CodexAgent) Version() (string, error)

Version returns the codex CLI version.

type CodexOption 

type CodexOption func(*CodexAgent)

CodexOption configures a CodexAgent.

func WithCodexBinaryPath 

func WithCodexBinaryPath(path string) CodexOption

WithCodexBinaryPath sets a custom path to the codex binary.

func WithCodexDefaultTimeout 

func WithCodexDefaultTimeout(d time.Duration) CodexOption

WithCodexDefaultTimeout sets the default execution timeout.

func WithCodexRunner 

func WithCodexRunner(r CommandRunner) CodexOption

WithCodexRunner sets a custom command runner (for testing).

func WithDangerouslyBypassApprovalsAndSandbox 

func WithDangerouslyBypassApprovalsAndSandbox(enabled bool) CodexOption

WithDangerouslyBypassApprovalsAndSandbox sets whether to pass --dangerously-bypass-approvals-and-sandbox.

type CommandRunner 

type CommandRunner interface {
	Run(ctx context.Context, name string, args []string, dir string, stdin string) (stdout, stderr string, exitCode int, err error)
}

CommandRunner executes shell commands. Allows mocking in tests.

type CopilotAgent added in v0.3.3 

type CopilotAgent struct {
	// contains filtered or unexported fields
}

CopilotAgent spawns GitHub Copilot CLI for task execution.

GitHub Copilot CLI implementation notes: - Supports both 'gh copilot' (passthrough to copilot binary) and standalone 'copilot' binary - Non-interactive mode: use -p/--prompt flag (exits after completion) - Uses --no-ask-user to disable the ask_user tool (fully autonomous) - Uses --silent to output only the agent response (no stats)

Install options: - Via gh: gh copilot (downloads copilot binary automatically if not in PATH) - Standalone: npm install -g @github/copilot or curl script - Usage: copilot -p "<prompt>" --no-ask-user --silent

func NewCopilotAgent added in v0.3.3 

func NewCopilotAgent(opts ...CopilotOption) *CopilotAgent

NewCopilotAgent creates a GitHub Copilot CLI agent.

func (*CopilotAgent) Available added in v0.3.3 

func (a *CopilotAgent) Available() bool

Available checks if the gh binary is available in PATH and copilot extension is installed.

func (*CopilotAgent) Execute added in v0.3.3 

func (a *CopilotAgent) Execute(ctx context.Context, opts ExecuteOptions) (*ExecuteResult, error)

Execute runs the Copilot CLI with the given prompt in non-interactive mode.

Both 'gh copilot' and standalone 'copilot' use the same -p flag interface. For 'gh copilot', '--' is used to pass flags through to the copilot binary.

func (*CopilotAgent) ExecuteWithFiles added in v0.3.3 

func (a *CopilotAgent) ExecuteWithFiles(ctx context.Context, prompt string, files []string, workDir string) (*ExecuteResult, error)

ExecuteWithFiles runs gh copilot with file context included.

func (*CopilotAgent) Name added in v0.3.3 

func (a *CopilotAgent) Name() string

Name returns "copilot".

func (*CopilotAgent) Version added in v0.3.3 

func (a *CopilotAgent) Version() (string, error)

Version returns the copilot CLI version.

type CopilotOption added in v0.3.3 

type CopilotOption func(*CopilotAgent)

CopilotOption configures a CopilotAgent.

func WithCopilotBinaryPath added in v0.3.3 

func WithCopilotBinaryPath(path string) CopilotOption

WithCopilotBinaryPath sets a custom path to the copilot binary ("gh" or "copilot").

func WithCopilotDangerouslySkipPermissions added in v0.3.4 

func WithCopilotDangerouslySkipPermissions(enabled bool) CopilotOption

WithCopilotDangerouslySkipPermissions sets whether to pass --allow-all-tools and --allow-all-urls.

func WithCopilotDefaultTimeout added in v0.3.3 

func WithCopilotDefaultTimeout(d time.Duration) CopilotOption

WithCopilotDefaultTimeout sets the default execution timeout.

func WithCopilotRunner added in v0.3.3 

func WithCopilotRunner(r CommandRunner) CopilotOption

WithCopilotRunner sets a custom command runner (for testing).

type ExecRunner 

type ExecRunner struct{}

ExecRunner is the default CommandRunner using os/exec.

func (*ExecRunner) Run 

func (r *ExecRunner) Run(ctx context.Context, name string, args []string, dir string, stdin string) (string, string, int, error)

Run executes a command and returns output.

type ExecuteOptions 

type ExecuteOptions struct {
	Prompt  string        // The prompt/task for the agent
	WorkDir string        // Working directory for execution
	Files   []string      // Optional file paths to include as context
	Timeout time.Duration // Execution timeout (0 = default)
}

ExecuteOptions configures an agent execution.

type ExecuteResult 

type ExecuteResult struct {
	Output   string        // Agent's text output
	JSON     []byte        // Structured JSON output if available
	ExitCode int           // Process exit code
	Duration time.Duration // Execution duration
	Error    string        // Error message if failed
}

ExecuteResult holds the outcome of an agent execution.

func (*ExecuteResult) IsSuccess 

func (r *ExecuteResult) IsSuccess() bool

IsSuccess returns true if the execution succeeded.

Source Files

View all Source files

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.