catence 0.2.0-beta.1 → 0.2.0-beta.10

package/scripts/discover-opencode-go.mjs

@@ -0,0 +1,210 @@
+// Discover OpenCode Go models and emit Catence Console profiles for them.
+//
+// OpenCode Go serves its catalog through three APIs: most models speak the
+// OpenAI Chat Completions API, Grok and GPT-5.6 Luna speak the OpenAI
+// Responses API, and MiniMax and Qwen speak the Anthropic Messages API. The
+// public `GET /models` endpoint returns only model ids, not which API each
+// model needs, so routing comes from the trusted table below and is intersected
+// with the live list (the same "live discovery + trusted metadata" approach
+// OpenClaw uses).
+//
+// Usage:
+//   node scripts/discover-opencode-go.mjs [--base-url <url>] [--write <config.json>] [--set-default]
+//
+// Flags:
+//   --base-url <url>   Root of the OpenCode Go API (default https://opencode.ai/zen/go/v1).
+//   --write <path>     Merge the two profiles into an existing config.json instead of printing to stdout.
+//   --set-default      With --write, set console.defaultProfile to opencode-go even when one already exists.
+
+import process from 'node:process';
+import { readFile, writeFile } from 'node:fs/promises';
+
+const DEFAULT_BASE_URL = 'https://opencode.ai/zen/go/v1';
+const API_KEY_ENV = 'OPENCODE_GO_API_KEY';
+const API_BASE_ENV = 'OPENCODE_GO_API_BASE';
+const MESSAGES_BASE_ENV = 'OPENCODE_GO_MESSAGES_API_BASE';
+
+// LiteLLM appends a fixed suffix to the profile base URL per provider, so the
+// three APIs need different prefixes and two base URLs:
+//   openai/<id>             -> {base}/chat/completions
+//   openai/responses/<id>   -> {base}/responses
+//   anthropic/<id>          -> {base minus /v1}/v1/messages
+const RESPONSES_MODELS = new Set(['grok-4.5', 'gpt-5.6-luna']);
+const MESSAGES_MODELS = new Set([
+  'minimax-m2.5',
+  'minimax-m2.7',
+  'minimax-m3',
+  'qwen3.5-plus',
+  'qwen3.6-plus',
+  'qwen3.7-plus',
+  'qwen3.7-max',
+  'qwen3.8-max',
+]);
+
+const BRAND_NAMES = new Map([
+  ['deepseek', 'DeepSeek'],
+  ['glm', 'GLM'],
+  ['gpt', 'GPT'],
+  ['grok', 'Grok'],
+  ['hy3', 'Hy3'],
+  ['kimi', 'Kimi'],
+  ['mimo', 'MiMo'],
+  ['minimax', 'MiniMax'],
+  ['qwen', 'Qwen'],
+]);
+
+function classify(id) {
+  if (RESPONSES_MODELS.has(id)) return { route: 'responses', guessed: false };
+  if (MESSAGES_MODELS.has(id)) return { route: 'messages', guessed: false };
+  if (/^(grok-|gpt-)/.test(id)) return { route: 'responses', guessed: true };
+  if (/^(minimax-|qwen)/.test(id)) return { route: 'messages', guessed: true };
+  return { route: 'chat', guessed: false };
+}
+
+function modelReference(id, route) {
+  if (route === 'responses') return `openai/responses/${id}`;
+  if (route === 'messages') return `anthropic/${id}`;
+  return `openai/${id}`;
+}
+
+function labelFor(id) {
+  return id
+    .split('-')
+    .map((segment) => BRAND_NAMES.get(segment) ?? segment.charAt(0).toUpperCase() + segment.slice(1))
+    .join(' ');
+}
+
+function usage() {
+  process.stderr.write(
+    'Usage: node scripts/discover-opencode-go.mjs [--base-url <url>] [--write <config.json>] [--set-default]\n',
+  );
+}
+
+async function main() {
+  const args = process.argv.slice(2);
+  let baseUrl = DEFAULT_BASE_URL;
+  let writePath = null;
+  let setDefault = false;
+  for (let i = 0; i < args.length; i += 1) {
+    const arg = args[i];
+    if (arg === '--base-url') {
+      baseUrl = args[i + 1];
+      if (!baseUrl) throw new Error('--base-url requires a URL.');
+      i += 1;
+    } else if (arg === '--write') {
+      writePath = args[i + 1];
+      if (!writePath) throw new Error('--write requires a config.json path.');
+      i += 1;
+    } else if (arg === '--set-default') {
+      setDefault = true;
+    } else if (arg === '--help' || arg === '-h') {
+      usage();
+      return;
+    } else {
+      throw new Error(`Unknown argument: ${arg}`);
+    }
+  }
+
+  if (!baseUrl.endsWith('/v1')) {
+    process.stderr.write(`Warning: --base-url ${baseUrl} does not end in /v1; chat and responses URLs may be wrong.\n`);
+  }
+  const modelsUrl = `${baseUrl.replace(/\/$/, '')}/models`;
+  const messagesBase = baseUrl.replace(/\/v1\/?$/, '');
+
+  let response;
+  try {
+    response = await fetch(modelsUrl, { headers: { accept: 'application/json' } });
+  } catch (error) {
+    throw new Error(`Could not reach ${modelsUrl}: ${error.message}`);
+  }
+  if (!response.ok) {
+    throw new Error(`GET ${modelsUrl} returned HTTP ${response.status}.`);
+  }
+  const payload = await response.json();
+  const ids = (payload.data ?? []).map((entry) => entry.id).filter((id) => typeof id === 'string' && id);
+
+  const groups = { chat: [], responses: [], messages: [] };
+  const guessed = [];
+  for (const id of ids) {
+    const { route, guessed: wasGuessed } = classify(id);
+    if (!(route in groups)) continue;
+    groups[route].push(id);
+    if (wasGuessed) guessed.push(`${id} -> ${route}`);
+  }
+  for (const route of Object.keys(groups)) groups[route].sort();
+
+  const chatProfileModels = {};
+  for (const id of groups.chat) {
+    chatProfileModels[id] = { label: labelFor(id), model: modelReference(id, 'chat') };
+  }
+  for (const id of groups.responses) {
+    chatProfileModels[id] = { label: labelFor(id), model: modelReference(id, 'responses') };
+  }
+  const messagesProfileModels = {};
+  for (const id of groups.messages) {
+    messagesProfileModels[id] = { label: labelFor(id), model: modelReference(id, 'messages') };
+  }
+
+  const chatDefault = groups.chat[0] ?? groups.responses[0];
+  const profiles = {};
+  profiles['opencode-go'] = {
+    label: 'OpenCode Go',
+    ...(chatDefault ? { defaultModel: chatDefault } : {}),
+    models: chatProfileModels,
+    apiKeyEnv: API_KEY_ENV,
+    apiBaseEnv: API_BASE_ENV,
+  };
+  if (messagesProfileModels && groups.messages.length) {
+    profiles['opencode-go-messages'] = {
+      label: 'OpenCode Go (Anthropic)',
+      defaultModel: groups.messages[0],
+      models: messagesProfileModels,
+      apiKeyEnv: API_KEY_ENV,
+      apiBaseEnv: MESSAGES_BASE_ENV,
+    };
+  }
+
+  const consoleSection = { defaultProfile: 'opencode-go', profiles };
+
+  if (guessed.length) {
+    process.stderr.write('Routed by prefix heuristic (verify before use):\n');
+    for (const line of guessed) process.stderr.write(`  ${line}\n`);
+  }
+  process.stderr.write(
+    `Discovered ${ids.length} models (${groups.chat.length} chat, ${groups.responses.length} responses, ${groups.messages.length} messages).\n`,
+  );
+
+  if (writePath) {
+    let root = {};
+    try {
+      root = JSON.parse(await readFile(writePath, 'utf8'));
+    } catch (error) {
+      if (error.code !== 'ENOENT') throw new Error(`Could not read ${writePath}: ${error.message}`);
+    }
+    if (typeof root !== 'object' || root === null || Array.isArray(root)) {
+      throw new Error(`${writePath} must contain a JSON object.`);
+    }
+    const existingConsole = typeof root.console === 'object' && root.console !== null ? root.console : {};
+    const mergedProfiles = { ...(existingConsole.profiles ?? {}), ...profiles };
+    const defaultProfile =
+      existingConsole.defaultProfile && !setDefault ? existingConsole.defaultProfile : 'opencode-go';
+    root.console = { ...existingConsole, defaultProfile, profiles: mergedProfiles };
+    await writeFile(writePath, `${JSON.stringify(root, null, 2)}\n`);
+    process.stderr.write(`Merged profiles into ${writePath} (defaultProfile: ${defaultProfile}).\n`);
+    return;
+  }
+
+  process.stderr.write(
+    `Export the key and base URLs, then add the JSON below under "console" in config.json:\n` +
+      `  export ${API_KEY_ENV}='sk-…'\n` +
+      `  export ${API_BASE_ENV}='${baseUrl}'\n` +
+      `  export ${MESSAGES_BASE_ENV}='${messagesBase}'\n` +
+      `(or rerun with --write ~/.catence/config.json to merge into an existing config.)\n`,
+  );
+  process.stdout.write(`${JSON.stringify(consoleSection, null, 2)}\n`);
+}
+
+await main().catch((error) => {
+  process.stderr.write(`Error: ${error.message}\n`);
+  process.exit(1);
+});

package/scripts/discover-opencode-zen.mjs

@@ -0,0 +1,218 @@
+// Discover OpenCode Zen models and emit Catence Console profiles for them.
+//
+// OpenCode Zen serves its catalog through `GET /v1/models` at
+// `https://opencode.ai/zen/v1`. Models are split between the chat/completions
+// endpoint and the responses endpoint. Routing is determined by model ID prefix:
+// GPT-5.x models use the responses API; everything else uses chat/completions.
+//
+// Usage:
+//   node scripts/discover-opencode-zen.mjs [--base-url <url>] [--write <config.json>] [--set-default] [--free-only]
+//
+// Flags:
+//   --base-url <url>   Root of the OpenCode Zen API (default https://opencode.ai/zen/v1).
+//   --write <path>     Merge the two profiles into an existing config.json instead of printing to stdout.
+//   --set-default      With --write, set console.defaultProfile to opencode-zen even when one already exists.
+//   --free-only        Only include free models (those with -free suffix or known free model IDs).
+
+import process from 'node:process';
+import { readFile, writeFile } from 'node:fs/promises';
+
+const DEFAULT_BASE_URL = 'https://opencode.ai/zen/v1';
+const API_KEY_ENV = 'OPENCODE_API_KEY';
+const API_BASE_ENV = 'OPENCODE_ZEN_API_BASE';
+const RESPONSES_BASE_ENV = 'OPENCODE_ZEN_RESPONSES_API_BASE';
+
+// Models that use the OpenAI Responses API (gpt-5.x series)
+const RESPONSES_PREFIXES = ['gpt-5', 'gpt-5.'];
+
+// Known free model IDs (in case the API doesn't mark them explicitly)
+const KNOWN_FREE_MODELS = new Set([
+  'deepseek-v4-flash-free',
+  'mimo-v2.5-free',
+  'nemotron-3-ultra-free',
+  'nemotron-3.5-lightning-free',
+  'hy3-free',
+  'laguna-s-2.1-free',
+  'big-pickle',
+  'north-mini-code-free',
+  'longcat-2.0-free',
+  'ling-3.0-flash-free',
+  'ling-3.0-tiny-free',
+]);
+
+const BRAND_NAMES = new Map([
+  ['deepseek', 'DeepSeek'],
+  ['gpt', 'GPT'],
+  ['mimo', 'MiMo'],
+  ['nemotron', 'Nemotron'],
+  ['hy3', 'Hy3'],
+  ['laguna', 'Laguna'],
+  ['big', 'Big'],
+  ['pickle', 'Pickle'],
+  ['north', 'North'],
+  ['mini', 'Mini'],
+  ['code', 'Code'],
+  ['longcat', 'LongCat'],
+  ['ling', 'Ling'],
+]);
+
+function isFreeModel(id) {
+  return id.endsWith('-free') || KNOWN_FREE_MODELS.has(id);
+}
+
+function classify(id) {
+  for (const prefix of RESPONSES_PREFIXES) {
+    if (id.startsWith(prefix)) return { route: 'responses', guessed: false };
+  }
+  return { route: 'chat', guessed: false };
+}
+
+function modelReference(id, route) {
+  if (route === 'responses') return `openai/responses/${id}`;
+  return `openai/${id}`;
+}
+
+function labelFor(id) {
+  return id
+    .split('-')
+    .map((segment) => BRAND_NAMES.get(segment) ?? segment.charAt(0).toUpperCase() + segment.slice(1))
+    .join(' ');
+}
+
+function usage() {
+  process.stderr.write(
+    'Usage: node scripts/discover-opencode-zen.mjs [--base-url <url>] [--write <config.json>] [--set-default] [--free-only]\n',
+  );
+}
+
+async function main() {
+  const args = process.argv.slice(2);
+  let baseUrl = DEFAULT_BASE_URL;
+  let writePath = null;
+  let setDefault = false;
+  let freeOnly = false;
+
+  for (let i = 0; i < args.length; i += 1) {
+    const arg = args[i];
+    if (arg === '--base-url') {
+      baseUrl = args[i + 1];
+      if (!baseUrl) throw new Error('--base-url requires a URL.');
+      i += 1;
+    } else if (arg === '--write') {
+      writePath = args[i + 1];
+      if (!writePath) throw new Error('--write requires a config.json path.');
+      i += 1;
+    } else if (arg === '--set-default') {
+      setDefault = true;
+    } else if (arg === '--free-only') {
+      freeOnly = true;
+    } else if (arg === '--help' || arg === '-h') {
+      usage();
+      return;
+    } else {
+      throw new Error(`Unknown argument: ${arg}`);
+    }
+  }
+
+  if (!baseUrl.endsWith('/v1')) {
+    process.stderr.write(`Warning: --base-url ${baseUrl} does not end in /v1; chat and responses URLs may be wrong.\n`);
+  }
+  const modelsUrl = `${baseUrl.replace(/\/$/, '')}/models`;
+
+  let response;
+  try {
+    response = await fetch(modelsUrl, { headers: { accept: 'application/json' } });
+  } catch (error) {
+    throw new Error(`Could not reach ${modelsUrl}: ${error.message}`);
+  }
+  if (!response.ok) {
+    throw new Error(`GET ${modelsUrl} returned HTTP ${response.status}.`);
+  }
+  const payload = await response.json();
+  const ids = (payload.data ?? [])
+    .map((entry) => entry.id)
+    .filter((id) => typeof id === 'string' && id);
+
+  const filteredIds = freeOnly ? ids.filter(isFreeModel) : ids;
+
+  const groups = { chat: [], responses: [] };
+  for (const id of filteredIds) {
+    const { route } = classify(id);
+    if (!(route in groups)) continue;
+    groups[route].push(id);
+  }
+  for (const route of Object.keys(groups)) groups[route].sort();
+
+  const chatProfileModels = {};
+  for (const id of groups.chat) {
+    chatProfileModels[id] = { label: labelFor(id), model: modelReference(id, 'chat') };
+  }
+  const responsesProfileModels = {};
+  for (const id of groups.responses) {
+    responsesProfileModels[id] = { label: labelFor(id), model: modelReference(id, 'responses') };
+  }
+
+  const chatDefault = groups.chat[0];
+  const responsesDefault = groups.responses[0];
+
+  const profiles = {};
+  if (groups.chat.length) {
+    profiles['opencode-zen'] = {
+      label: 'OpenCode Zen',
+      ...(chatDefault ? { defaultModel: chatDefault } : {}),
+      models: chatProfileModels,
+      apiKeyEnv: API_KEY_ENV,
+      apiBaseEnv: API_BASE_ENV,
+    };
+  }
+  if (groups.responses.length) {
+    profiles['opencode-zen-responses'] = {
+      label: 'OpenCode Zen (Responses)',
+      ...(responsesDefault ? { defaultModel: responsesDefault } : {}),
+      models: responsesProfileModels,
+      apiKeyEnv: API_KEY_ENV,
+      apiBaseEnv: RESPONSES_BASE_ENV,
+    };
+  }
+
+  const consoleSection = { defaultProfile: 'opencode-zen', profiles };
+
+  process.stderr.write(
+    `Discovered ${ids.length} models (${groups.chat.length} chat, ${groups.responses.length} responses).` +
+    (freeOnly ? ` Filtered to ${filteredIds.length} free models.` : '') + '\n',
+  );
+
+  if (writePath) {
+    let root = {};
+    try {
+      root = JSON.parse(await readFile(writePath, 'utf8'));
+    } catch (error) {
+      if (error.code !== 'ENOENT') throw new Error(`Could not read ${writePath}: ${error.message}`);
+    }
+    if (typeof root !== 'object' || root === null || Array.isArray(root)) {
+      throw new Error(`${writePath} must contain a JSON object.`);
+    }
+    const existingConsole = typeof root.console === 'object' && root.console !== null ? root.console : {};
+    const mergedProfiles = { ...(existingConsole.profiles ?? {}), ...profiles };
+    const defaultProfile =
+      existingConsole.defaultProfile && !setDefault ? existingConsole.defaultProfile : 'opencode-zen';
+    root.console = { ...existingConsole, defaultProfile, profiles: mergedProfiles };
+    await writeFile(writePath, `${JSON.stringify(root, null, 2)}\n`);
+    process.stderr.write(`Merged profiles into ${writePath} (defaultProfile: ${defaultProfile}).\n`);
+    return;
+  }
+
+  process.stderr.write(
+    `Export the key and base URLs, then add the JSON below under "console" in config.json:\n` +
+      `  export ${API_KEY_ENV}='sk-…'\n` +
+      `  export ${API_BASE_ENV}='${baseUrl}'\n` +
+      `  export ${RESPONSES_BASE_ENV}='${baseUrl}'\n` +
+      `(or rerun with --write ~/.catence/config.json to merge into an existing config.)\n`,
+  );
+  process.stdout.write(`${JSON.stringify(consoleSection, null, 2)}\n`);
+}
+
+await main().catch((error) => {
+  process.stderr.write(`Error: ${error.message}\n`);
+  process.exit(1);
+});