@myvillage/cli 1.5.0 → 1.5.1

package/src/index.js

@@ -53,6 +53,18 @@ import {
   bizreqsStatusCommand,
   bizreqsImportCommand,
 } from './commands/bizreqs.js';
+import {
+  soulprintInitCommand,
+  soulprintIngestCommand,
+  soulprintDatasetListCommand,
+  soulprintDatasetPullCommand,
+  soulprintTrainCommand,
+  soulprintJobsCommand,
+  soulprintJobDetailCommand,
+  soulprintPushCommand,
+  soulprintPublishCommand,
+  soulprintModelsCommand,
+} from './commands/soulprint.js';
 
 const require = createRequire(import.meta.url);
 const pkg = require('../package.json');
@@ -73,6 +85,7 @@ export function run() {
   program
     .command('login')
     .description('Authenticate with MyVillageOS')
+    .option('--no-browser', 'Manual login for headless environments (servers, SSH, etc.)')
     .action(loginCommand);
 
   program
@@ -357,5 +370,105 @@ export function run() {
     .option('--contact <name>', 'Contact name')
     .action(bizreqsImportCommand);
 
+  // ── SoulPrint Studio: Model Training Pipeline ───────────
+
+  const soulprintCmd = program
+    .command('soulprint')
+    .description('SoulPrint Studio \u2014 datasets, training, and model publishing');
+
+  soulprintCmd
+    .command('init')
+    .description('Initialize local training workspace')
+    .option('--skip-python', 'Skip Python venv setup')
+    .option('--skip-scripts', 'Skip training script download')
+    .action(soulprintInitCommand);
+
+  soulprintCmd
+    .command('ingest <path>')
+    .description('Ingest training data into a dataset')
+    .requiredOption('--dataset <slug>', 'Target dataset slug')
+    .requiredOption('--type <type>', 'Data type: text, image, audio, structured, multimodal')
+    .option('--source <source>', 'Ingestion source label', 'CLI')
+    .option('--captions <path>', 'Path to captions file (image/multimodal)')
+    .option('--transcriptions <path>', 'Path to transcriptions file (audio)')
+    .option('--schema <path>', 'Path to schema JSON file (structured)')
+    .option('--recursive', 'Recurse into subdirectories')
+    .option('--glob <pattern>', 'File glob pattern to filter')
+    .option('--split <split>', 'Assign all items to a split: train, validation, test')
+    .option('--dry-run', 'Show what would be ingested without uploading')
+    .option('--concurrency <n>', 'Parallel upload limit', '5')
+    .action(soulprintIngestCommand);
+
+  const soulprintDatasetCmd = soulprintCmd
+    .command('datasets')
+    .description('Manage training datasets');
+
+  soulprintDatasetCmd
+    .command('list')
+    .description('List available datasets on SoulPrint Studio')
+    .option('--type <type>', 'Filter by type: text, image, audio, structured, multimodal')
+    .option('--status <status>', 'Filter by status: collecting, ready, training, archived')
+    .option('--json', 'Output raw JSON')
+    .action(soulprintDatasetListCommand);
+
+  soulprintDatasetCmd
+    .command('pull <slug>')
+    .description('Download a dataset to your local machine')
+    .option('--version <number>', 'Specific version (default: latest frozen)')
+    .option('--split <split>', 'Download only a specific split: train, validation, test')
+    .option('--force', 'Re-download even if already present locally')
+    .action(soulprintDatasetPullCommand);
+
+  soulprintCmd
+    .command('train')
+    .description('Run model training locally')
+    .requiredOption('--type <type>', 'Training type: text, image, audio, structured, multimodal')
+    .requiredOption('--dataset <slug>', 'Dataset slug to train on')
+    .option('--version <number>', 'Dataset version (default: latest)')
+    .option('--base <model>', 'Base model identifier')
+    .option('--method <method>', 'Training method')
+    .option('--config <path>', 'Path to training config YAML file')
+    .option('--name <name>', 'Name for the resulting model')
+    .option('--dry-run', 'Validate config and dataset without starting training')
+    .action(soulprintTrainCommand);
+
+  soulprintCmd
+    .command('jobs [jobId]')
+    .description('View training job status')
+    .option('--status <status>', 'Filter by status')
+    .option('--json', 'Output raw JSON')
+    .action((jobId, options) => {
+      if (jobId) return soulprintJobDetailCommand(jobId, options);
+      return soulprintJobsCommand(options);
+    });
+
+  soulprintCmd
+    .command('models')
+    .description('List models in the registry')
+    .option('--status <status>', 'Filter: draft, validated, published, rejected')
+    .option('--type <type>', 'Filter: text, image, audio, structured, multimodal')
+    .option('--json', 'Output raw JSON')
+    .action(soulprintModelsCommand);
+
+  soulprintCmd
+    .command('push <path>')
+    .description('Upload locally trained model artifacts to SoulPrint Studio')
+    .requiredOption('--name <name>', 'Model name')
+    .requiredOption('--type <type>', 'Model type: text, image, audio, structured, multimodal')
+    .option('--base <model>', 'Base model it was trained from')
+    .option('--method <method>', 'Training method used')
+    .option('--job <jobId>', 'Link to an existing training job')
+    .option('--description <text>', 'Model description')
+    .action(soulprintPushCommand);
+
+  soulprintCmd
+    .command('publish <modelSlug>')
+    .description('Publish a validated model to the MyVillage platform')
+    .option('--villager <id>', 'Target villager ID to own the model')
+    .option('--villages <ids>', 'Comma-separated village IDs to associate')
+    .option('--tier <tier>', 'Model tier: FREE, BASIC, PRO, ENTERPRISE', 'FREE')
+    .option('--public', 'Make the model publicly available')
+    .action(soulprintPublishCommand);
+
   program.parse();
 }

package/src/utils/agent-scaffolder.js

@@ -5,9 +5,9 @@ import { stringify as stringifyYaml } from 'yaml';
 // ── MCP Tool Catalog ────────────────────────────────────
 
 const TOOL_CATALOG = {
-  'man-feed': {
-    command: 'internal',
-    description: 'MAN feed access (post, read, react)',
+  myvillage: {
+    url: 'https://mcp.myvillageproject.ai',
+    description: 'MyVillageOS platform access (feed, posts, communities, wallet, knowledge)',
     always_enabled: true,
   },
   filesystem: {
@@ -151,11 +151,11 @@ ${description}
 function generateToolsYaml(selectedTools) {
   const servers = {};
 
-  // man-feed is always included
-  servers['man-feed'] = TOOL_CATALOG['man-feed'];
+  // myvillage is always included
+  servers['myvillage'] = TOOL_CATALOG['myvillage'];
 
   for (const toolId of selectedTools) {
-    if (toolId === 'man-feed') continue;
+    if (toolId === 'myvillage') continue;
     if (TOOL_CATALOG[toolId]) {
       servers[toolId] = { ...TOOL_CATALOG[toolId] };
     }

package/src/utils/api.js

@@ -7,7 +7,7 @@ const require = createRequire(import.meta.url);
 const { version } = require('../../package.json');
 const USER_AGENT = `MyVillageOS-CLI/${version}`;
 
-function createClient(baseURL) {
+export function createClient(baseURL) {
   const client = axios.create({
     baseURL,
     headers: {

package/src/utils/config.js

@@ -10,6 +10,7 @@ const DEFAULT_CONFIG = {
   networkBaseUrl: 'https://portal.myvillageproject.ai/api/network',
   bizreqsBaseUrl: 'https://portal.myvillageproject.ai/api/bizreqs',
   oauthBaseUrl: 'https://portal.myvillageproject.ai/api/oauth',
+  soulprintBaseUrl: 'https://soulprint-studio.myvillageproject.ai/api',
   clientId: 'mvos_aG_c729fuQxvvqYHOnkgTQ',
   callbackPort: 3737,
   anthropicApiKey: null,

package/src/utils/soulprint-api.js

@@ -0,0 +1,136 @@
+import { createClient } from './api.js';
+import { getConfig } from './config.js';
+
+// ── Client Factory ─────────────────────────────────────
+
+export function getSoulprintClient() {
+  const config = getConfig();
+  return createClient(config.soulprintBaseUrl);
+}
+
+// ── Datasets ───────────────────────────────────────────
+
+export async function listDatasets(params = {}) {
+  const client = getSoulprintClient();
+  const response = await client.get('/datasets', { params });
+  return response.data;
+}
+
+export async function getDataset(slug) {
+  const client = getSoulprintClient();
+  const response = await client.get(`/datasets/${slug}`);
+  return response.data;
+}
+
+export async function getDatasetDownload(slug, version) {
+  const client = getSoulprintClient();
+  const params = version ? { version } : {};
+  const response = await client.get(`/datasets/${slug}/download`, { params });
+  return response.data;
+}
+
+// ── Training Jobs ──────────────────────────────────────
+
+export async function createJob(data) {
+  const client = getSoulprintClient();
+  const response = await client.post('/jobs', data);
+  return response.data;
+}
+
+export async function listJobs(params = {}) {
+  const client = getSoulprintClient();
+  const response = await client.get('/jobs', { params });
+  return response.data;
+}
+
+export async function getJob(id) {
+  const client = getSoulprintClient();
+  const response = await client.get(`/jobs/${id}`);
+  return response.data;
+}
+
+export async function updateJobStatus(id, data) {
+  const client = getSoulprintClient();
+  const response = await client.patch(`/jobs/${id}/status`, data);
+  return response.data;
+}
+
+export async function completeJob(id, data) {
+  const client = getSoulprintClient();
+  const response = await client.post(`/jobs/${id}/complete`, data);
+  return response.data;
+}
+
+export async function failJob(id, data) {
+  const client = getSoulprintClient();
+  const response = await client.post(`/jobs/${id}/fail`, data);
+  return response.data;
+}
+
+// ── Models ─────────────────────────────────────────────
+
+export async function listModels(params = {}) {
+  const client = getSoulprintClient();
+  const response = await client.get('/models', { params });
+  return response.data;
+}
+
+export async function getModel(slug) {
+  const client = getSoulprintClient();
+  const response = await client.get(`/models/${slug}`);
+  return response.data;
+}
+
+export async function publishModel(slug, data) {
+  const client = getSoulprintClient();
+  const response = await client.post(`/models/${slug}/publish`, data);
+  return response.data;
+}
+
+// ── Ingestion ──────────────────────────────────────────
+
+export async function ingestText(data) {
+  const client = getSoulprintClient();
+  const response = await client.post('/ingest/text', data);
+  return response.data;
+}
+
+export async function ingestStructured(data) {
+  const client = getSoulprintClient();
+  const response = await client.post('/ingest/structured', data);
+  return response.data;
+}
+
+export async function prepareIngestion(data) {
+  const client = getSoulprintClient();
+  const response = await client.post('/ingest/prepare', data);
+  return response.data;
+}
+
+export async function completeIngestion(ingestionId, data) {
+  const client = getSoulprintClient();
+  const response = await client.post(`/ingest/${ingestionId}/complete`, data);
+  return response.data;
+}
+
+export async function abortIngestion(ingestionId) {
+  const client = getSoulprintClient();
+  const response = await client.post(`/ingest/${ingestionId}/abort`);
+  return response.data;
+}
+
+// ── Upload ─────────────────────────────────────────────
+
+export async function getUploadUrls(data) {
+  const client = getSoulprintClient();
+  const response = await client.post('/upload/presign', data);
+  return response.data;
+}
+
+// ── Scripts ────────────────────────────────────────────
+
+export async function getScriptsManifest() {
+  const client = getSoulprintClient();
+  const response = await client.get('/scripts/manifest');
+  return response.data;
+}

package/src/utils/soulprint-workspace.js

@@ -0,0 +1,158 @@
+import { existsSync, mkdirSync, readFileSync, writeFileSync, readdirSync } from 'fs';
+import { join } from 'path';
+import { execSync } from 'child_process';
+import { getConfigDir } from './config.js';
+import { parse as parseYaml, stringify as stringifyYaml } from 'yaml';
+
+const SOULPRINT_DIR_NAME = 'soulprint';
+
+// ── Directory Helpers ──────────────────────────────────
+
+export function getSoulprintDir() {
+  return join(getConfigDir(), SOULPRINT_DIR_NAME);
+}
+
+export function getDatasetsDir() {
+  return join(getSoulprintDir(), 'datasets');
+}
+
+export function getModelsDir() {
+  return join(getSoulprintDir(), 'models');
+}
+
+export function getConfigsDir() {
+  return join(getSoulprintDir(), 'configs');
+}
+
+export function getLogsDir() {
+  return join(getSoulprintDir(), 'logs');
+}
+
+export function getVenvDir() {
+  return join(getSoulprintDir(), 'venv');
+}
+
+export function getScriptsDir() {
+  return join(getSoulprintDir(), 'scripts');
+}
+
+// ── Workspace State ────────────────────────────────────
+
+export function isWorkspaceInitialized() {
+  const dir = getSoulprintDir();
+  return existsSync(dir) && existsSync(join(dir, 'workspace.yaml'));
+}
+
+export function ensureWorkspace() {
+  return isWorkspaceInitialized();
+}
+
+export function readWorkspaceConfig() {
+  const configPath = join(getSoulprintDir(), 'workspace.yaml');
+  if (!existsSync(configPath)) return null;
+  try {
+    return parseYaml(readFileSync(configPath, 'utf-8'));
+  } catch {
+    return null;
+  }
+}
+
+export function writeWorkspaceConfig(config) {
+  const configPath = join(getSoulprintDir(), 'workspace.yaml');
+  writeFileSync(configPath, stringifyYaml(config, { lineWidth: 0 }));
+}
+
+// ── Dataset Helpers ────────────────────────────────────
+
+export function getLocalDatasetDir(slug, version) {
+  return join(getDatasetsDir(), slug, `v${version}`);
+}
+
+export function isDatasetDownloaded(slug, version) {
+  const dir = getLocalDatasetDir(slug, version);
+  return existsSync(dir) && existsSync(join(dir, 'manifest.json'));
+}
+
+export function listLocalDatasets() {
+  const dir = getDatasetsDir();
+  if (!existsSync(dir)) return [];
+  return readdirSync(dir, { withFileTypes: true })
+    .filter(e => e.isDirectory())
+    .map(e => e.name);
+}
+
+// ── Python Environment ─────────────────────────────────
+
+export function getPythonPath() {
+  const venvPython = join(getVenvDir(), 'bin', 'python');
+  if (existsSync(venvPython)) return venvPython;
+  return null;
+}
+
+export function isPythonAvailable() {
+  try {
+    execSync('python3 --version', { stdio: 'pipe' });
+    return true;
+  } catch {
+    return false;
+  }
+}
+
+export function getPythonVersion() {
+  try {
+    return execSync('python3 --version', { stdio: 'pipe' }).toString().trim();
+  } catch {
+    return null;
+  }
+}
+
+export function detectGPU() {
+  try {
+    const nvidiaSmi = execSync('nvidia-smi --query-gpu=name,memory.total --format=csv,noheader', { stdio: 'pipe' }).toString().trim();
+    if (nvidiaSmi) {
+      const [name, vram] = nvidiaSmi.split(', ');
+      return { type: 'cuda', name, vram };
+    }
+  } catch { /* no CUDA */ }
+
+  try {
+    const sysctl = execSync('sysctl -n machdep.cpu.brand_string', { stdio: 'pipe' }).toString().trim();
+    if (sysctl.includes('Apple')) {
+      const mem = execSync('sysctl -n hw.memsize', { stdio: 'pipe' }).toString().trim();
+      const memGB = Math.round(parseInt(mem) / 1073741824);
+      return { type: 'mps', name: sysctl, vram: `${memGB}GB unified` };
+    }
+  } catch { /* not Apple Silicon */ }
+
+  return { type: 'cpu', name: 'CPU only', vram: null };
+}
+
+export function getMachineInfo() {
+  const gpu = detectGPU();
+  return {
+    gpu: gpu.name,
+    gpuType: gpu.type,
+    vram: gpu.vram,
+    os: `${process.platform} ${process.arch}`,
+    python: getPythonVersion(),
+    nodeVersion: process.version,
+  };
+}
+
+// ── Training Job Local State ───────────────────────────
+
+export function getJobOutputDir(jobId) {
+  return join(getModelsDir(), jobId);
+}
+
+export function getJobLogFile(jobId) {
+  const dir = getLogsDir();
+  if (!existsSync(dir)) mkdirSync(dir, { recursive: true });
+  return join(dir, `${jobId}.jsonl`);
+}
+
+export function appendJobLog(jobId, entry) {
+  const logFile = getJobLogFile(jobId);
+  const line = JSON.stringify({ ...entry, ts: new Date().toISOString() }) + '\n';
+  writeFileSync(logFile, line, { flag: 'a' });
+}