Programmatic exports for spawning an AI agent with a prompt and for resolving which command and arguments to use. These mirror the behavior used by the AIDD agent CLI without going through the command line.
runAgent spawns an agent process with the given prompt. It returns a promise that resolves when the process exits with code 0 and rejects otherwise.
RunAgentOptions is { agentConfig: AgentConfig; prompt: string; cwd: string }, where AgentConfig is { command: string; args?: string[] }.
import { runAgent } from 'aidd/agent';
await runAgent({
agentConfig: { command: 'claude', args: ['-p'] },
prompt: 'Build a login page',
cwd: '/my-project',
});Resolve a built-in preset or full configuration from project settings and environment. Use getAgentConfig when you only need a named preset; use resolveAgentConfig when you want the same resolution order as the tooling (explicit value, then env, then aidd-custom/config.yml, then default).
getAgentConfig(name?: string): AgentConfig;import { getAgentConfig } from 'aidd/agent-config';
const config = getAgentConfig('claude');
// => { command: 'claude', args: ['-p'] }resolveAgentConfig(options?: {
value?: string | AgentConfig;
cwd?: string;
}): Promise<AgentConfig>;import { resolveAgentConfig } from 'aidd/agent-config';
// Priority: explicit value → AIDD_AGENT_CONFIG env → aidd-custom/config.yml → claude
const config = await resolveAgentConfig({ cwd: process.cwd() });For CLI usage, see Agent CLI.