@stackmemoryai/stackmemory
Version:
Project-scoped memory for AI coding tools. Durable context across sessions with MCP integration, frames, smart retrieval, Claude Code skills, and automatic hooks.
534 lines (532 loc) • 18.5 kB
JavaScript
import { fileURLToPath as __fileURLToPath } from 'url';
import { dirname as __pathDirname } from 'path';
const __filename = __fileURLToPath(import.meta.url);
const __dirname = __pathDirname(__filename);
import { Command } from "commander";
import chalk from "chalk";
import inquirer from "inquirer";
import {
loadModelRouterConfig,
saveModelRouterConfig,
getModelRouter,
getFallbackStatus,
triggerFallback,
resetFallback
} from "../../core/models/model-router.js";
function createModelCommand() {
const model = new Command("model").description(
"Configure model routing between Claude and alternative providers"
).addHelpText(
"after",
`
Examples:
stackmemory model status Show current model configuration
stackmemory model enable Enable model routing
stackmemory model disable Disable model routing (use Claude only)
stackmemory model route plan qwen Route plan tasks to Qwen
stackmemory model provider qwen Configure Qwen provider
stackmemory model thinking Configure thinking mode settings
`
);
model.command("status").description("Show current model router configuration").action(() => {
const config = loadModelRouterConfig();
const router = getModelRouter();
console.log(chalk.cyan("\nModel Router Status"));
console.log(chalk.gray("\u2500".repeat(40)));
console.log(
` Enabled: ${config.enabled ? chalk.green("Yes") : chalk.gray("No")}`
);
console.log(` Default Provider: ${chalk.white(config.defaultProvider)}`);
console.log(
` Current Provider: ${chalk.white(router.getCurrentProvider())}`
);
console.log(chalk.cyan("\nTask Routing"));
const routes = config.taskRouting;
console.log(` Plan tasks: ${routes.plan || chalk.gray("(default)")}`);
console.log(` Think tasks: ${routes.think || chalk.gray("(default)")}`);
console.log(` Code tasks: ${routes.code || chalk.gray("(default)")}`);
console.log(
` Review tasks: ${routes.review || chalk.gray("(default)")}`
);
console.log(chalk.cyan("\nConfigured Providers"));
for (const [name, provider] of Object.entries(config.providers)) {
if (provider) {
const hasKey = process.env[provider.apiKeyEnv] ? chalk.green("*") : chalk.red("!");
console.log(` ${hasKey} ${name}: ${provider.model}`);
if (provider.baseUrl) {
console.log(chalk.gray(` URL: ${provider.baseUrl}`));
}
}
}
console.log(chalk.cyan("\nThinking Mode"));
console.log(
` Enabled: ${config.thinkingMode.enabled ? chalk.green("Yes") : chalk.gray("No")}`
);
if (config.thinkingMode.budget) {
console.log(` Budget: ${config.thinkingMode.budget} tokens`);
}
if (config.thinkingMode.temperature) {
console.log(` Temperature: ${config.thinkingMode.temperature}`);
}
console.log(chalk.cyan("\nFallback (Auto)"));
const fallbackStatus = getFallbackStatus();
console.log(
` Enabled: ${fallbackStatus.enabled ? chalk.green("Yes") : chalk.gray("No")}`
);
if (fallbackStatus.enabled) {
const keyStatus = fallbackStatus.hasApiKey ? chalk.green("ready") : chalk.red("no API key");
console.log(` Provider: ${fallbackStatus.provider} (${keyStatus})`);
console.log(
` Triggers: ${[
config.fallback?.onRateLimit && "rate-limit",
config.fallback?.onError && "errors",
config.fallback?.onTimeout && "timeout"
].filter(Boolean).join(", ")}`
);
if (fallbackStatus.inFallback) {
console.log(
chalk.yellow(` Status: IN FALLBACK (${fallbackStatus.reason})`)
);
}
}
console.log(chalk.gray("\n* = API key found, ! = API key missing"));
});
model.command("enable").description("Enable model routing").action(() => {
const config = loadModelRouterConfig();
config.enabled = true;
saveModelRouterConfig(config);
console.log(chalk.green("[OK] Model routing enabled"));
});
model.command("disable").description("Disable model routing (use Claude only)").action(() => {
const config = loadModelRouterConfig();
config.enabled = false;
saveModelRouterConfig(config);
console.log(
chalk.green("[OK] Model routing disabled (using Claude only)")
);
});
model.command("route <task> [provider]").description("Route a task type to a specific provider").addHelpText(
"after",
`
Task types: plan, think, code, review
Providers: anthropic, qwen, openai, ollama, custom
Examples:
stackmemory model route plan qwen Route planning to Qwen
stackmemory model route think qwen Route deep thinking to Qwen
stackmemory model route plan Clear plan routing (use default)
`
).action((task, provider) => {
const validTasks = ["plan", "think", "code", "review"];
const validProviders = [
"anthropic",
"qwen",
"openai",
"ollama",
"custom"
];
if (!validTasks.includes(task)) {
console.error(
chalk.red(`Invalid task type: ${task}. Use: ${validTasks.join(", ")}`)
);
process.exit(1);
}
if (provider && !validProviders.includes(provider)) {
console.error(
chalk.red(
`Invalid provider: ${provider}. Use: ${validProviders.join(", ")}`
)
);
process.exit(1);
}
const config = loadModelRouterConfig();
if (provider) {
config.taskRouting[task] = provider;
saveModelRouterConfig(config);
console.log(chalk.green(`[OK] ${task} tasks routed to ${provider}`));
} else {
delete config.taskRouting[task];
saveModelRouterConfig(config);
console.log(
chalk.green(`[OK] ${task} routing cleared (using default)`)
);
}
});
model.command("provider <name>").description("Configure a model provider").option("-m, --model <model>", "Model name/ID").option("-u, --url <url>", "Base URL for API").option("-k, --key-env <env>", "Environment variable for API key").option("-i, --interactive", "Interactive configuration").action(async (name, options) => {
const validProviders = [
"anthropic",
"qwen",
"openai",
"ollama",
"custom"
];
if (!validProviders.includes(name)) {
console.error(
chalk.red(
`Invalid provider: ${name}. Use: ${validProviders.join(", ")}`
)
);
process.exit(1);
}
const config = loadModelRouterConfig();
if (options.interactive) {
const answers = await inquirer.prompt([
{
type: "input",
name: "model",
message: "Model name/ID:",
default: config.providers[name]?.model || getDefaultModel(name)
},
{
type: "input",
name: "baseUrl",
message: "Base URL (leave empty for provider default):",
default: config.providers[name]?.baseUrl || getDefaultUrl(name)
},
{
type: "input",
name: "apiKeyEnv",
message: "Environment variable for API key:",
default: config.providers[name]?.apiKeyEnv || getDefaultKeyEnv(name)
}
]);
const providerConfig = {
provider: name,
model: answers.model,
apiKeyEnv: answers.apiKeyEnv
};
if (answers.baseUrl) {
providerConfig.baseUrl = answers.baseUrl;
}
config.providers[name] = providerConfig;
} else {
const existing = config.providers[name] || {
provider: name,
model: getDefaultModel(name),
apiKeyEnv: getDefaultKeyEnv(name)
};
if (options.model) existing.model = options.model;
if (options.url) existing.baseUrl = options.url;
if (options.keyEnv) existing.apiKeyEnv = options.keyEnv;
config.providers[name] = existing;
}
saveModelRouterConfig(config);
console.log(chalk.green(`[OK] Provider ${name} configured`));
const provider = config.providers[name];
if (provider) {
console.log(chalk.gray(` Model: ${provider.model}`));
if (provider.baseUrl) {
console.log(chalk.gray(` URL: ${provider.baseUrl}`));
}
console.log(chalk.gray(` Key env: ${provider.apiKeyEnv}`));
const hasKey = process.env[provider.apiKeyEnv];
if (!hasKey) {
console.log(
chalk.yellow(
`
[WARN] ${provider.apiKeyEnv} not set in environment`
)
);
}
}
});
model.command("thinking").description("Configure thinking mode settings").option("--enable", "Enable thinking mode").option("--disable", "Disable thinking mode").option("-b, --budget <tokens>", "Max thinking tokens", parseInt).option("-t, --temperature <temp>", "Temperature (0.0-1.0)", parseFloat).option("-p, --top-p <topP>", "Top P (0.0-1.0)", parseFloat).action((options) => {
const config = loadModelRouterConfig();
if (options.enable !== void 0) {
config.thinkingMode.enabled = true;
}
if (options.disable !== void 0) {
config.thinkingMode.enabled = false;
}
if (options.budget !== void 0) {
config.thinkingMode.budget = options.budget;
}
if (options.temperature !== void 0) {
config.thinkingMode.temperature = options.temperature;
}
if (options.topP !== void 0) {
config.thinkingMode.topP = options.topP;
}
saveModelRouterConfig(config);
console.log(chalk.green("[OK] Thinking mode configured"));
console.log(chalk.gray(` Enabled: ${config.thinkingMode.enabled}`));
if (config.thinkingMode.budget) {
console.log(
chalk.gray(` Budget: ${config.thinkingMode.budget} tokens`)
);
}
if (config.thinkingMode.temperature) {
console.log(
chalk.gray(` Temperature: ${config.thinkingMode.temperature}`)
);
}
if (config.thinkingMode.topP) {
console.log(chalk.gray(` Top P: ${config.thinkingMode.topP}`));
}
});
model.command("setup-qwen").description("Quick setup for Qwen provider (DashScope)").action(async () => {
console.log(chalk.cyan("\nQwen Provider Setup"));
console.log(chalk.gray("\u2500".repeat(40)));
console.log(
chalk.gray(
"Qwen3-Max-Thinking supports extended reasoning with thinking mode."
)
);
console.log(
chalk.gray("API: https://dashscope.aliyuncs.com/compatible-mode/v1\n")
);
const answers = await inquirer.prompt([
{
type: "input",
name: "model",
message: "Model name:",
default: "qwen3-max-2025-01-23"
},
{
type: "confirm",
name: "enableThinking",
message: "Enable thinking mode?",
default: true
},
{
type: "number",
name: "thinkingBudget",
message: "Thinking budget (tokens):",
default: 1e4,
when: (a) => a.enableThinking
},
{
type: "confirm",
name: "routePlan",
message: "Route plan tasks to Qwen?",
default: true
},
{
type: "confirm",
name: "routeThink",
message: "Route think tasks to Qwen?",
default: true
}
]);
const config = loadModelRouterConfig();
config.providers.qwen = {
provider: "qwen",
model: answers.model,
baseUrl: "https://dashscope.aliyuncs.com/compatible-mode/v1",
apiKeyEnv: "DASHSCOPE_API_KEY",
params: {
enable_thinking: answers.enableThinking,
thinking_budget: answers.thinkingBudget || 1e4
}
};
if (answers.enableThinking) {
config.thinkingMode = {
enabled: true,
budget: answers.thinkingBudget || 1e4,
temperature: 0.6,
topP: 0.95
};
}
if (answers.routePlan) {
config.taskRouting.plan = "qwen";
}
if (answers.routeThink) {
config.taskRouting.think = "qwen";
}
config.enabled = true;
saveModelRouterConfig(config);
console.log(chalk.green("\n[OK] Qwen provider configured"));
console.log(chalk.gray(" Model: " + answers.model));
console.log(
chalk.gray(
" Thinking mode: " + (answers.enableThinking ? "Enabled" : "Disabled")
)
);
if (!process.env["DASHSCOPE_API_KEY"]) {
console.log(chalk.yellow("\n[WARN] DASHSCOPE_API_KEY not set"));
console.log(chalk.gray("Add to your environment:"));
console.log(chalk.white(" export DASHSCOPE_API_KEY=your-api-key"));
}
});
model.command("default <provider>").description("Set the default provider").action((provider) => {
const validProviders = [
"anthropic",
"qwen",
"openai",
"ollama",
"custom"
];
if (!validProviders.includes(provider)) {
console.error(
chalk.red(
`Invalid provider: ${provider}. Use: ${validProviders.join(", ")}`
)
);
process.exit(1);
}
const config = loadModelRouterConfig();
config.defaultProvider = provider;
saveModelRouterConfig(config);
console.log(chalk.green(`[OK] Default provider set to ${provider}`));
});
model.command("fallback").description("Configure automatic fallback to Qwen").option("--enable", "Enable automatic fallback").option("--disable", "Disable automatic fallback").option("-p, --provider <provider>", "Set fallback provider").option("--on-rate-limit", "Fallback on rate limit (429)").option("--no-rate-limit", "Disable rate limit fallback").option("--on-error", "Fallback on server errors (5xx)").option("--no-error", "Disable error fallback").option("--on-timeout", "Fallback on timeout").option("--no-timeout", "Disable timeout fallback").option("--test", "Test fallback by triggering it manually").option("--reset", "Reset fallback state (exit fallback mode)").action((options) => {
const config = loadModelRouterConfig();
if (!config.fallback) {
config.fallback = {
enabled: true,
provider: "qwen",
onRateLimit: true,
onError: true,
onTimeout: true,
maxRetries: 2,
retryDelayMs: 1e3
};
}
if (options.test) {
console.log(chalk.yellow("Testing fallback..."));
const env = triggerFallback("manual");
if (Object.keys(env).length > 0) {
console.log(chalk.green("[OK] Fallback activated"));
console.log(chalk.gray(` Provider: ${config.fallback.provider}`));
console.log(chalk.gray(` Model: ${env["ANTHROPIC_MODEL"]}`));
} else {
console.log(chalk.red("Fallback not available"));
console.log(chalk.gray(" Check: DASHSCOPE_API_KEY is set"));
}
return;
}
if (options.reset) {
resetFallback();
console.log(chalk.green("[OK] Fallback state reset"));
return;
}
let updated = false;
if (options.enable !== void 0) {
config.fallback.enabled = true;
updated = true;
}
if (options.disable !== void 0) {
config.fallback.enabled = false;
updated = true;
}
if (options.provider) {
config.fallback.provider = options.provider;
updated = true;
}
if (options.onRateLimit !== void 0) {
config.fallback.onRateLimit = true;
updated = true;
}
if (options.rateLimit === false) {
config.fallback.onRateLimit = false;
updated = true;
}
if (options.onError !== void 0) {
config.fallback.onError = true;
updated = true;
}
if (options.error === false) {
config.fallback.onError = false;
updated = true;
}
if (options.onTimeout !== void 0) {
config.fallback.onTimeout = true;
updated = true;
}
if (options.timeout === false) {
config.fallback.onTimeout = false;
updated = true;
}
if (updated) {
saveModelRouterConfig(config);
console.log(chalk.green("[OK] Fallback configuration updated"));
}
const status = getFallbackStatus();
console.log(chalk.cyan("\nFallback Configuration"));
console.log(chalk.gray("\u2500".repeat(30)));
console.log(
` Enabled: ${status.enabled ? chalk.green("Yes") : chalk.gray("No")}`
);
console.log(` Provider: ${config.fallback.provider}`);
console.log(
` API Key: ${status.hasApiKey ? chalk.green("Set") : chalk.red("Missing")}`
);
console.log(
` On Rate Limit: ${config.fallback.onRateLimit ? "Yes" : "No"}`
);
console.log(` On Error: ${config.fallback.onError ? "Yes" : "No"}`);
console.log(` On Timeout: ${config.fallback.onTimeout ? "Yes" : "No"}`);
if (status.inFallback) {
console.log(
chalk.yellow(`
Currently IN FALLBACK (${status.reason})`)
);
}
if (!status.hasApiKey) {
console.log(chalk.yellow("\nTo enable Qwen fallback:"));
console.log(chalk.gray(" export DASHSCOPE_API_KEY=your-api-key"));
}
});
model.command("reset").description("Reset model router to defaults").action(async () => {
const { confirmed } = await inquirer.prompt([
{
type: "confirm",
name: "confirmed",
message: "Reset model router configuration to defaults?",
default: false
}
]);
if (confirmed) {
const defaultConfig = {
enabled: false,
defaultProvider: "anthropic",
taskRouting: {},
providers: {
anthropic: {
provider: "anthropic",
model: "claude-sonnet-4-20250514",
apiKeyEnv: "ANTHROPIC_API_KEY"
}
},
thinkingMode: {
enabled: false
}
};
saveModelRouterConfig(defaultConfig);
console.log(chalk.green("[OK] Model router reset to defaults"));
} else {
console.log(chalk.gray("Reset cancelled"));
}
});
return model;
}
function getDefaultModel(provider) {
const defaults = {
anthropic: "claude-sonnet-4-20250514",
qwen: "qwen3-max-2025-01-23",
openai: "gpt-4o",
ollama: "llama3.2",
custom: "custom-model"
};
return defaults[provider] || "unknown";
}
function getDefaultUrl(provider) {
const defaults = {
qwen: "https://dashscope.aliyuncs.com/compatible-mode/v1",
ollama: "http://localhost:11434/v1"
};
return defaults[provider] || "";
}
function getDefaultKeyEnv(provider) {
const defaults = {
anthropic: "ANTHROPIC_API_KEY",
qwen: "DASHSCOPE_API_KEY",
openai: "OPENAI_API_KEY",
ollama: "OLLAMA_API_KEY",
custom: "CUSTOM_API_KEY"
};
return defaults[provider] || "API_KEY";
}
export {
createModelCommand
};
//# sourceMappingURL=model.js.map