gpteam 0.1.26 → 0.1.28

package/README.md

@@ -12,7 +12,9 @@ When a selected client is missing, the interactive CLI shows the exact install c
 
 Recommendation order is deterministic: success rate first, then an experience score. The score weighs first SSE event time, total completion time, p90 completion tail latency, and health-check time, so a node with one very slow probe is not recommended just because its median result looks good. The result table also marks the fastest full completion separately from the balanced recommendation, because deep or long-output model runs may care more about full completion time than first-token responsiveness. Model prompts use "configurable context" wording because the value is written to the client-side Codex `model_context_window`; it must not be confused with a public marketing total-window label.
 
-Client config writing follows the same safety pattern as cc-switch: keep Codex top-level fields separate from provider tables, preserve unrelated sections such as MCP servers, merge OpenCode/OpenClaw providers additively, and stop before writing when an existing JSON/JSON5 config cannot be parsed. GPTeam keeps three proxy-specific extensions on top of that baseline: Codex explicitly disables WebSocket prewarm, Codex/OpenCode/Claude Code write the `gpteam_image` MCP server so Image 2 can be called from chat, and Claude Code writes the reasoning-effort header through `ANTHROPIC_CUSTOM_HEADERS`.
+Production endpoints are shown as the main, Japan, and Hong Kong ingress domains. They are all GPTeam production API entry points backed by the current new backend stack; the CLI no longer describes them as split-routing or secondary-worker nodes. The current node ids are `main`, `jp`, and `hk`. Older scripted values `jp-direct`, `jp-split`, and `hk-split` remain accepted as compatibility aliases.
+
+Client config writing follows the same safety pattern as cc-switch: keep Codex top-level fields separate from provider tables, preserve unrelated sections such as MCP servers, merge OpenCode/OpenClaw providers additively, and stop before writing when an existing JSON/JSON5 config cannot be parsed. GPTeam keeps three proxy-specific extensions on top of that baseline: Codex enables WebSocket capability for the GPTeam provider, Codex/OpenCode/Claude Code write the `gpteam_image` MCP server so Image 2 can be called from chat, and Claude Code writes the reasoning-effort header through `ANTHROPIC_CUSTOM_HEADERS`.
 
 The Image MCP config uses cc-switch-style per-client env blocks. Codex writes `[mcp_servers.gpteam_image.env]`, OpenCode writes `mcp.gpteam_image.environment`, and Claude Code writes `mcpServers.gpteam_image.env` in `~/.claude.json`. The MCP receives `GPTEAM_API_KEY` and `GPTEAM_BASE_URL` from that MCP config, so it does not depend on Codex `auth.json` or inherited `OPENAI_API_KEY`.
 
@@ -21,7 +23,7 @@ The Image MCP exposes an async-first local job flow plus a legacy compatibility
 - `create_image_job`: recommended for normal use. It starts a local background image job and returns `job_id` quickly, which avoids losing the whole generation when a VPN, proxy, or client has a 60-second idle timeout.
 - `get_image_job_status`: checks whether the local job is queued, running, succeeded, failed, canceled, or expired.
 - `cancel_image_job`: cancels a queued/running local job.
-- `download_image_result`: returns the completed file metadata and local file path by default. Pass `include_image: true` only when the client must receive MCP image content, because large images can trigger context compaction.
+- `download_image_result`: returns the completed file metadata and image content. Use `metadata_only`, `include_image`, and `include_revised_prompt` to control large result payloads.
 - `get_capabilities`: returns supported sizes, formats, quality values, async support, cancellation semantics, queue limits, and image-to-image support.
 - `generate_image`: legacy compatibility alias. It now creates the same async job and returns `job_id` immediately instead of blocking until the image is complete.
 

package/lib/bench.js

@@ -41,6 +41,7 @@ export async function benchmarkNode(node, options) {
 export function summarizeNode(node, samples) {
   const successful = samples.filter((sample) => sample.ok);
   const streams = successful.map((sample) => sample.stream);
+  const allStreams = samples.map((sample) => sample.stream).filter(Boolean);
   const firstEvents = streams.map((item) => item.firstEventMs).filter(Number.isFinite);
   const totals = streams.map((item) => item.totalMs).filter(Number.isFinite);
   const summary = {
@@ -51,10 +52,10 @@ export function summarizeNode(node, samples) {
     totalMs: median(totals),
     tailFirstEventMs: percentile(firstEvents, 0.9),
     tailTotalMs: percentile(totals, 0.9),
-    dnsMs: median(streams.map((item) => item.dnsMs).filter(Number.isFinite)),
-    tcpMs: median(streams.map((item) => item.tcpMs).filter(Number.isFinite)),
-    tlsMs: median(streams.map((item) => item.tlsMs).filter(Number.isFinite)),
-    healthMs: median(successful.map((sample) => sample.health.totalMs).filter(Number.isFinite)),
+    dnsMs: median(allStreams.map((item) => item.dnsMs).filter(Number.isFinite)),
+    tcpMs: median(allStreams.map((item) => item.tcpMs).filter(Number.isFinite)),
+    tlsMs: median(allStreams.map((item) => item.tlsMs).filter(Number.isFinite)),
+    healthMs: median(samples.map((sample) => sample.health?.totalMs).filter(Number.isFinite)),
     error: samples.find((sample) => sample.error)?.error || ''
   };
   summary.experienceScore = scoreResult(summary);
@@ -157,16 +158,17 @@ function measureStream(baseUrl, options) {
       });
       response.on('end', () => {
         const semantic = inspectSSEBody(body);
-        const ok = response.statusCode >= 200
-          && response.statusCode < 300
+        const status = response.statusCode || 0;
+        const ok = status >= 200
+          && status < 300
           && semantic.ok
           && Number.isFinite(timings.firstEventMs);
         resolve({
           ok,
-          status: response.statusCode,
+          status,
           ...timings,
           totalMs: performance.now() - started,
-          error: ok ? '' : semantic.error || `stream HTTP ${response.statusCode}`
+          error: ok ? '' : formatStreamProbeError(status, response.headers, body, semantic.error)
         });
       });
     });
@@ -194,6 +196,40 @@ function measureStream(baseUrl, options) {
   });
 }
 
+export function formatStreamProbeError(status, headers, body, semanticError) {
+  const statusCode = Number(status || 0);
+  const contentType = String(headers?.['content-type'] || headers?.['Content-Type'] || '').trim();
+  const detail = responseBodyDetail(body);
+  const parts = [];
+  if (statusCode < 200 || statusCode >= 300) {
+    parts.push(`stream HTTP ${statusCode}`);
+  } else if (semanticError === 'stream empty response' && detail) {
+    parts.push('stream non-SSE response');
+  } else {
+    parts.push(semanticError || `stream HTTP ${statusCode}`);
+  }
+  if (contentType) parts.push(`content-type ${contentType}`);
+  if (detail) parts.push(detail);
+  return parts.join('；');
+}
+
+function responseBodyDetail(body) {
+  const text = String(body || '').replace(/\s+/g, ' ').trim();
+  if (!text) return '';
+  const parsed = parseJSON(text);
+  const message = parsed?.error?.message || parsed?.message || parsed?.error;
+  if (message) return `body ${String(message).slice(0, 180)}`;
+  return `body ${text.slice(0, 180)}`;
+}
+
+function parseJSON(value) {
+  try {
+    return JSON.parse(value);
+  } catch {
+    return null;
+  }
+}
+
 function median(values) {
   if (!values.length) return NaN;
   const sorted = [...values].sort((a, b) => a - b);

package/lib/cli.js

@@ -5,7 +5,7 @@ import { ensureClientInstalled, formatInstallCommand } from './client-install.js
 import { CLIENTS, writeClientConfig } from './config.js';
 import { getHelpText, PACKAGE_NAME, PACKAGE_VERSION } from './help.js';
 import { modelByID, validateApiKey } from './models.js';
-import { describeSplit, INGRESS_NODES } from './nodes.js';
+import { INGRESS_NODES, nodeMatchesID } from './nodes.js';
 import { createTheme, stripAnsi } from './terminal.js';
 
 export async function runCli(argv = []) {
@@ -104,7 +104,7 @@ export function printResults(results, theme = createTheme()) {
   const recommended = results.find((item) => item.successRate > 0);
   const fastestTotal = findFastestTotalResult(results);
   const rows = results.map((item) => [
-    `${item.node.label}（${describeSplit(item.node)}）`,
+    formatNodeLabel(item.node),
     formatMs(item.dnsMs),
     formatMs(item.tcpMs),
     formatMs(item.tlsMs),
@@ -164,10 +164,11 @@ async function askContextLength(rl, model, preferred) {
 
 export async function chooseNode(rl, results, preferred, recommendedID, theme = createTheme()) {
   const nodes = results.map((item) => item.node);
-  const preferredNode = nodes.find((node) => node.id === preferred);
+  const preferredNode = nodes.find((node) => nodeMatchesID(node, preferred));
   if (preferredNode) return preferredNode;
   if (recommendedID) {
-    printStatus(theme, '综合推荐', recommendedID);
+    const recommendedNode = nodes.find((node) => nodeMatchesID(node, recommendedID));
+    printStatus(theme, '综合推荐', recommendedNode ? formatNodeLabel(recommendedNode) : recommendedID);
   }
   return (await choose(rl, '请选择最终写入的入口', nodes.map((node) => ({
     id: node.id,
@@ -223,7 +224,7 @@ export function formatModelLabel(model) {
 }
 
 export function formatNodeLabel(node) {
-  return `${node.label}（${describeSplit(node)}）`;
+  return String(node?.label || node?.id || '');
 }
 
 export function assertNodeConfig(node) {

package/lib/config.js

@@ -54,7 +54,7 @@ export function writeCodexConfig(settings) {
     `base_url = ${tomlString(settings.node.baseUrl)}`,
     'wire_api = "responses"',
     'requires_openai_auth = true',
-    'supports_websockets = false'
+    'supports_websockets = true'
   ];
   const mcpCommand = codexImageMCPCommand();
   const managedImageMCP = [

package/lib/help.js

@@ -1,5 +1,5 @@
 export const PACKAGE_NAME = 'gpteam';
-export const PACKAGE_VERSION = '0.1.26';
+export const PACKAGE_VERSION = '0.1.28';
 
 export function getHelpText() {
   return [
@@ -15,7 +15,7 @@ export function getHelpText() {
     '  --model <id>             预选模型，例如 gpt-5.5',
     '  --context <tokens>       预设可配置上下文长度',
     '  --effort <level>         按模型支持项预选，常见为 none / low / medium / high / xhigh',
-    '  --node <id>              jp-direct / jp-split / hk-split / us-split',
+    '  --node <id>              main / jp / hk（兼容旧参数 jp-direct / jp-split / hk-split）',
     '  --rounds <n>             每个入口测速轮数，默认 3',
     '  --max-output-tokens <n>  测速输出上限，默认 648',
     '  --install-client         客户端命令缺失时直接安装，不再二次确认',

package/lib/image-mcp/image.js

@@ -651,7 +651,7 @@ function normalizeIdempotencyKey(value) {
 }
 
 function shapeDownloadResult(result, input = {}) {
-  const includeImage = input.include_image === true && !input.metadata_only;
+  const includeImage = !input.metadata_only && input.include_image !== false;
   const includeRevisedPrompt = input.include_revised_prompt !== false;
   const output = {
     ...result,

package/lib/image-mcp/server.js

@@ -238,13 +238,13 @@ function downloadSchema() {
   const schema = jobIDSchema();
   schema.properties.metadata_only = {
     type: 'boolean',
-    description: '只返回文件路径和元数据，不返回 MCP 图片内容。默认 true，避免大图进入上下文触发频繁 compact。',
-    default: true
+    description: 'Return only metadata without MCP image content.',
+    default: false
   };
   schema.properties.include_image = {
     type: 'boolean',
-    description: '显式返回 MCP 图片内容。大图会显著增加上下文，通常保持 false，只使用本地文件路径。',
-    default: false
+    description: 'Include image content when the job succeeded.',
+    default: true
   };
   schema.properties.include_revised_prompt = {
     type: 'boolean',

package/lib/nodes.js

@@ -1,40 +1,31 @@
 export const INGRESS_NODES = [
   {
-    id: 'jp-direct',
-    label: '直连',
-    region: 'JP',
-    split: false,
+    id: 'main',
+    aliases: ['jp-direct'],
+    label: '主入口',
     baseUrl: 'https://api.gpteamservices.com/v1',
     healthUrl: 'https://api.gpteamservices.com/api/health'
   },
   {
-    id: 'jp-split',
+    id: 'jp',
+    aliases: ['jp-split'],
     label: '日本入口',
-    region: 'JP',
-    split: true,
     baseUrl: 'https://api-jp.gpteamservices.com/v1',
     healthUrl: 'https://api-jp.gpteamservices.com/api/health'
   },
   {
-    id: 'hk-split',
+    id: 'hk',
+    aliases: ['hk-split'],
     label: '香港入口',
-    region: 'HK',
-    split: true,
     baseUrl: 'https://api-hk.gpteamservices.com/v1',
     healthUrl: 'https://api-hk.gpteamservices.com/api/health'
-  },
-  {
-    id: 'us-split',
-    label: '美国入口',
-    region: 'US',
-    split: true,
-    baseUrl: 'https://api-us.gpteamservices.com/v1',
-    healthUrl: 'https://api-us.gpteamservices.com/api/health'
   }
 ];
 
-export function describeSplit(node) {
-  return node.split ? '分流' : '不分流';
+export function nodeMatchesID(node, id) {
+  const value = String(id || '').trim();
+  if (!value) return false;
+  return node.id === value || (Array.isArray(node.aliases) && node.aliases.includes(value));
 }
 
 export function endpointRoot(baseUrl) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "gpteam",
-  "version": "0.1.26",
+  "version": "0.1.28",
   "description": "GPTeam API interactive client configurator and ingress benchmark CLI.",
   "type": "module",
   "bin": {