@lightcone-ai/daemon 0.15.53 → 0.15.54

package/mcp-servers/_thin-proxy/forward.js

@@ -0,0 +1,80 @@
+// Generic thin-proxy helper for MCP servers whose implementation lives on
+// the lightcone server (D17 anti-leak). Each daemon-side MCP server using
+// this helper just declares its tool list and a server_id; per-tool calls
+// are forwarded to /internal/agent/:agentId/mcp/:serverId/:toolName.
+//
+// Usage (in daemon/mcp-servers/<some-name>/index.js):
+//
+//   import { startThinProxy } from '../../_thin-proxy/forward.js';
+//   await startThinProxy({
+//     serverId: 'platform-policy-db',
+//     serverName: 'official-platform-policy-db',
+//     tools: [
+//       { name: 'platform_policy_db', description: '...', inputSchema: { ... } },
+//     ],
+//   });
+
+import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
+import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js';
+import { z } from 'zod';
+
+const SERVER_URL = process.env.SERVER_URL ?? '';
+const MACHINE_API_KEY = process.env.MACHINE_API_KEY ?? '';
+const AGENT_ID = process.env.AGENT_ID ?? '';
+
+function toTextContent(payload) {
+  let text;
+  try {
+    text = typeof payload === 'string' ? payload : JSON.stringify(payload, null, 2);
+  } catch {
+    text = String(payload);
+  }
+  return { content: [{ type: 'text', text }] };
+}
+
+async function forwardToServer(serverId, toolName, args) {
+  if (!SERVER_URL || !MACHINE_API_KEY || !AGENT_ID) {
+    throw new Error('thin-proxy missing SERVER_URL / MACHINE_API_KEY / AGENT_ID env');
+  }
+  const url = `${SERVER_URL}/internal/agent/${encodeURIComponent(AGENT_ID)}/mcp/${encodeURIComponent(serverId)}/${encodeURIComponent(toolName)}`;
+  const res = await fetch(url, {
+    method: 'POST',
+    headers: {
+      'Content-Type': 'application/json',
+      'Authorization': `Bearer ${MACHINE_API_KEY}`,
+    },
+    body: JSON.stringify(args ?? {}),
+  });
+  let body = null;
+  try { body = await res.json(); } catch { /* ignore */ }
+  if (!res.ok || !body?.ok) {
+    const reason = body?.error ?? `HTTP ${res.status}`;
+    throw new Error(reason);
+  }
+  return body.result;
+}
+
+export async function startThinProxy({ serverId, serverName, version = '1.0.0', tools }) {
+  const server = new McpServer({ name: serverName ?? serverId, version });
+  for (const tool of tools ?? []) {
+    if (!tool?.name) continue;
+    const description = tool.description ?? '';
+    // We intentionally accept ANY input here — the server-side handler
+    // performs full validation, so the proxy stays thin and free of
+    // duplicated schema definitions.
+    const inputSchema = tool.inputSchema ?? { args: z.any().optional() };
+    server.tool(tool.name, description, inputSchema, async (args) => {
+      try {
+        const result = await forwardToServer(serverId, tool.name, args ?? {});
+        return toTextContent(result);
+      } catch (err) {
+        return {
+          isError: true,
+          content: [{ type: 'text', text: `Error: ${err?.message ?? String(err)}` }],
+        };
+      }
+    });
+  }
+  await server.connect(new StdioServerTransport());
+  console.error(`[${serverId}] thin-proxy started → ${SERVER_URL || '<no-server-url>'}`);
+}

package/mcp-servers/official/audience-research/index.js

@@ -1,383 +1,31 @@
 #!/usr/bin/env node
+// Thin-proxy. Real impl on lightcone server.
+// Source of truth: src/mcp-services/audience-research.js
 import { z } from 'zod';
-import { startFixtureServer } from '../../official-common/server.js';
+import { startThinProxy } from '../../_thin-proxy/forward.js';
 
-const FIXTURE_META = Object.freeze({
-  mode: 'fixture',
-  as_of: '2026-04-29',
-  capability: 'audience-research',
-  disclaimer: 'Audience profile is estimated from fixture fan snapshots + public benchmark assumptions, not live platform BI exports.',
-});
-
-const PLATFORM_ALIAS = Object.freeze({
-  xiaohongshu: 'xhs',
-  redbook: 'xhs',
-  douyin: 'douyin',
-  tiktok_cn: 'douyin',
-  wechat: 'wechat-mp',
-  wechat_mp: 'wechat-mp',
-  wechatmp: 'wechat-mp',
-  gzh: 'wechat-mp',
-  '\u516c\u4f17\u53f7': 'wechat-mp',
-});
-
-const PLATFORM_BENCHMARKS = Object.freeze({
-  xhs: Object.freeze({
-    default_audience: {
-      age_range: '25-34',
-      gender_skew: '\u5973\u6027\u4e3b\u5bfc',
-      geo_focus: ['\u4e0a\u6d77', '\u676d\u5dde', '\u6df1\u5733', '\u5317\u4eac'],
-      occupation_tags: ['\u767d\u9886', '\u81ea\u7531\u804c\u4e1a', '\u5b66\u751f'],
-      pain_points: ['\u9009\u9898\u540c\u8d28\u5316', '\u7b14\u8bb0\u8f6c\u5316\u4f4e', '\u62cd\u6444\u6267\u884c\u6210\u672c\u9ad8'],
-      aspirations: ['\u7a33\u5b9a\u6da8\u7c89', '\u5efa\u7acb\u4e13\u4e1a\u4eba\u8bbe', '\u83b7\u5f97\u54c1\u724c\u5408\u4f5c'],
-    },
-    pools: {
-      cities: ['\u4e0a\u6d77', '\u676d\u5dde', '\u82cf\u5dde', '\u6df1\u5733', '\u6210\u90fd', '\u5317\u4eac', '\u5e7f\u5dde'],
-      occupations: ['\u767d\u9886', '\u5185\u5bb9\u8fd0\u8425', '\u8bbe\u8ba1\u5e08', '\u5b66\u751f', '\u81ea\u7531\u804c\u4e1a'],
-      pain_points: ['\u9009\u9898\u540c\u8d28\u5316', '\u5728\u7ebf\u7d22\u5c11', '\u66dd\u5149\u4e0d\u7a33\u5b9a', '\u62cd\u6444\u6548\u7387\u4f4e'],
-      aspirations: ['\u589e\u52a0\u54c1\u724c\u4fe1\u4efb', '\u63d0\u9ad8\u79c1\u57df\u6210\u4ea4', '\u5f62\u6210\u56fa\u5b9a\u4eba\u8bbe', '\u5185\u5bb9\u53ef\u590d\u7528'],
-      female_ratio: 0.68,
-      age_peak: 29,
-    },
-  }),
-  douyin: Object.freeze({
-    default_audience: {
-      age_range: '18-34',
-      gender_skew: '\u5747\u8861',
-      geo_focus: ['\u5317\u4eac', '\u5e7f\u5dde', '\u6210\u90fd', '\u6b66\u6c49'],
-      occupation_tags: ['\u5e74\u8f7b\u767d\u9886', '\u5b66\u751f', '\u4e2a\u4f53\u7ecf\u8425\u8005'],
-      pain_points: ['\u5b8c\u64ad\u7387\u4e0d\u7a33\u5b9a', '\u811a\u672c\u751f\u786c', '\u8d77\u53f7\u6210\u672c\u9ad8'],
-      aspirations: ['\u7206\u6b3e\u7387\u63d0\u5347', '\u4f18\u5316\u8d26\u53f7\u6807\u7b7e', '\u83b7\u5f97\u5546\u5355\u7ebf\u7d22'],
-    },
-    pools: {
-      cities: ['\u5317\u4eac', '\u5e7f\u5dde', '\u6df1\u5733', '\u91cd\u5e86', '\u6210\u90fd', '\u897f\u5b89', '\u957f\u6c99'],
-      occupations: ['\u5b66\u751f', '\u9500\u552e', '\u5e74\u8f7b\u767d\u9886', '\u4e3b\u64ad', '\u4e2a\u4f53\u7ecf\u8425\u8005'],
-      pain_points: ['\u5b8c\u64ad\u7387\u4e0d\u7a33\u5b9a', '\u8bc4\u8bba\u8f6c\u5316\u4f4e', '\u7d20\u6750\u7f3a\u4e4f', '\u66dd\u5149\u6ce2\u52a8\u5927'],
-      aspirations: ['\u63d0\u9ad8\u81ea\u7136\u6d41\u91cf', '\u589e\u52a0\u76f4\u64ad\u8fdb\u623f', '\u7a33\u5b9a\u7c89\u4e1d\u7c98\u6027', '\u63d0\u5347\u8d26\u53f7\u6743\u91cd'],
-      female_ratio: 0.52,
-      age_peak: 27,
-    },
-  }),
-  'wechat-mp': Object.freeze({
-    default_audience: {
-      age_range: '25-44',
-      gender_skew: '\u5747\u8861',
-      geo_focus: ['\u5317\u4eac', '\u4e0a\u6d77', '\u5e7f\u5dde', '\u6df1\u5733'],
-      occupation_tags: ['\u804c\u573a\u7ba1\u7406\u8005', '\u4ea7\u54c1\u7ecf\u7406', '\u4e2d\u5c0f\u4f01\u4e1a\u4e3b'],
-      pain_points: ['\u9605\u8bfb\u6253\u5f00\u7387\u4e0b\u6ed1', '\u6587\u7ae0\u5b8c\u8bfb\u7387\u4f4e', '\u79c1\u57df\u8f6c\u5316\u65ad\u5c42'],
-      aspirations: ['\u5efa\u7acb\u7a33\u5b9a\u793e\u7fa4', '\u63d0\u9ad8\u590d\u8d2d\u7387', '\u63d0\u9ad8\u4e13\u4e1a\u5f71\u54cd\u529b'],
-    },
-    pools: {
-      cities: ['\u5317\u4eac', '\u4e0a\u6d77', '\u6df1\u5733', '\u5e7f\u5dde', '\u676d\u5dde', '\u6b66\u6c49'],
-      occupations: ['\u804c\u573a\u7ba1\u7406\u8005', '\u4ea7\u54c1\u7ecf\u7406', '\u4f01\u4e1a\u670d\u52a1', '\u8fd0\u8425\u8d1f\u8d23\u4eba', '\u4e2d\u5c0f\u4f01\u4e1a\u4e3b'],
-      pain_points: ['\u9605\u8bfb\u6253\u5f00\u7387\u4e0b\u6ed1', '\u9009\u9898\u8001\u5316', '\u9500\u552e\u627f\u63a5\u65ad\u5c42', '\u66f4\u65b0\u8282\u594f\u4e0d\u7a33\u5b9a'],
-      aspirations: ['\u63d0\u5347\u4e13\u4e1a\u8ba4\u77e5', '\u589e\u5f3a\u79c1\u57df\u8f6c\u5316', '\u5b9e\u73b0\u8fde\u8f7d\u5316\u8f93\u51fa', '\u5f62\u6210\u9ad8\u4ef7\u503c\u54a8\u8be2\u7ebf\u7d22'],
-      female_ratio: 0.5,
-      age_peak: 33,
-    },
-  }),
-});
-
-const CSV_FIELD_ALIASES = Object.freeze({
-  age: 'age',
-  age_value: 'age',
-  age_range: 'age',
-  gender: 'gender',
-  sex: 'gender',
-  city: 'city',
-  geo: 'city',
-  location: 'city',
-  occupation: 'occupation',
-  job: 'occupation',
-  pain_point: 'pain_point',
-  painpoint: 'pain_point',
-  challenge: 'pain_point',
-  aspiration: 'aspiration',
-  goal: 'aspiration',
-});
-
-function normalizePlatform(value, accountId = '') {
-  const raw = String(value ?? '').trim().toLowerCase();
-  if (raw) return PLATFORM_ALIAS[raw] ?? raw;
-
-  const id = String(accountId ?? '').toLowerCase();
-  if (id.includes('douyin')) return 'douyin';
-  if (id.includes('wechat') || id.includes('gzh') || id.includes('mp')) return 'wechat-mp';
-  return 'xhs';
-}
-
-function clampInt(value, min, max, fallback) {
-  const n = Number(value);
-  if (!Number.isFinite(n)) return fallback;
-  return Math.max(min, Math.min(max, Math.trunc(n)));
-}
-
-function deterministicRatio(seed, salt) {
-  const text = `${seed}#${salt}`;
-  let hash = 2166136261;
-  for (let i = 0; i < text.length; i += 1) {
-    hash ^= text.charCodeAt(i);
-    hash = Math.imul(hash, 16777619);
-  }
-  return (hash >>> 0) / 4294967295;
-}
-
-function parseAge(value) {
-  const text = String(value ?? '').trim();
-  if (!text) return null;
-  const num = Number(text);
-  if (Number.isFinite(num)) return Math.max(13, Math.min(70, Math.round(num)));
-
-  const match = text.match(/(\d{2})/);
-  if (!match) return null;
-  return Math.max(13, Math.min(70, Number(match[1])));
-}
-
-function normalizeGender(value) {
-  const raw = String(value ?? '').trim().toLowerCase();
-  if (!raw) return 'unknown';
-  if (raw === 'f' || raw.includes('female') || raw.includes('\u5973')) return 'female';
-  if (raw === 'm' || raw.includes('male') || raw.includes('\u7537')) return 'male';
-  return 'unknown';
-}
-
-function normalizeToken(value) {
-  const text = String(value ?? '').trim();
-  return text || null;
-}
-
-function normalizeSampleRow(row = {}) {
-  const normalized = {
-    age: null,
-    gender: 'unknown',
-    city: null,
-    occupation: null,
-    pain_point: null,
-    aspiration: null,
-  };
-
-  for (const [rawKey, rawValue] of Object.entries(row)) {
-    const key = CSV_FIELD_ALIASES[String(rawKey ?? '').trim().toLowerCase()] ?? null;
-    if (!key) continue;
-
-    if (key === 'age') normalized.age = parseAge(rawValue);
-    else if (key === 'gender') normalized.gender = normalizeGender(rawValue);
-    else normalized[key] = normalizeToken(rawValue);
-  }
-
-  return normalized;
-}
-
-function parseCsv(text, maxRows) {
-  const raw = String(text ?? '').trim();
-  if (!raw) return [];
-
-  const lines = raw
-    .split(/\r?\n/)
-    .map((line) => line.trim())
-    .filter(Boolean);
-  if (lines.length < 2) return [];
-
-  const headers = lines[0].split(',').map((item) => item.trim());
-  const rows = [];
-
-  for (const line of lines.slice(1)) {
-    if (rows.length >= maxRows) break;
-    const cols = line.split(',').map((item) => item.trim());
-    const row = {};
-    for (let i = 0; i < headers.length; i += 1) {
-      row[headers[i]] = cols[i] ?? '';
-    }
-    rows.push(normalizeSampleRow(row));
-  }
-
-  return rows;
-}
-
-function generateSyntheticSamples({ accountId, platform, count }) {
-  const benchmark = PLATFORM_BENCHMARKS[platform] ?? PLATFORM_BENCHMARKS.xhs;
-  const pools = benchmark.pools;
-
-  const samples = [];
-  for (let i = 0; i < count; i += 1) {
-    const ageNoise = deterministicRatio(accountId, `age:${i}`) * 14 - 7;
-    const age = Math.max(16, Math.min(55, Math.round(pools.age_peak + ageNoise)));
-    const isFemale = deterministicRatio(accountId, `gender:${i}`) < pools.female_ratio;
-
-    samples.push({
-      age,
-      gender: isFemale ? 'female' : 'male',
-      city: pools.cities[Math.floor(deterministicRatio(accountId, `city:${i}`) * pools.cities.length)],
-      occupation: pools.occupations[Math.floor(deterministicRatio(accountId, `job:${i}`) * pools.occupations.length)],
-      pain_point: pools.pain_points[Math.floor(deterministicRatio(accountId, `pain:${i}`) * pools.pain_points.length)],
-      aspiration: pools.aspirations[Math.floor(deterministicRatio(accountId, `asp:${i}`) * pools.aspirations.length)],
-    });
-  }
-
-  return samples;
-}
-
-function ageBucket(age) {
-  if (!Number.isFinite(age)) return null;
-  if (age <= 24) return '18-24';
-  if (age <= 34) return '25-34';
-  if (age <= 44) return '35-44';
-  return '45+';
-}
-
-function topItems(samples, key, limit = 5) {
-  const counter = new Map();
-  for (const sample of samples) {
-    const value = normalizeToken(sample?.[key]);
-    if (!value) continue;
-    counter.set(value, (counter.get(value) ?? 0) + 1);
-  }
-
-  return [...counter.entries()]
-    .sort((a, b) => {
-      if (b[1] !== a[1]) return b[1] - a[1];
-      return a[0].localeCompare(b[0], 'zh-Hans-CN');
-    })
-    .slice(0, limit)
-    .map((item) => item[0]);
-}
-
-function mergeDistinct(primary, fallback, limit = 5) {
-  const output = [];
-  const seen = new Set();
-
-  for (const value of [...primary, ...fallback]) {
-    const text = String(value ?? '').trim();
-    if (!text || seen.has(text)) continue;
-    seen.add(text);
-    output.push(text);
-    if (output.length >= limit) break;
-  }
-
-  return output;
-}
-
-function inferAudience({ samples, benchmark }) {
-  const ageCounter = new Map();
-  let female = 0;
-  let male = 0;
-
-  for (const sample of samples) {
-    const bucket = ageBucket(sample.age);
-    if (bucket) ageCounter.set(bucket, (ageCounter.get(bucket) ?? 0) + 1);
-
-    if (sample.gender === 'female') female += 1;
-    else if (sample.gender === 'male') male += 1;
-  }
-
-  const topAgeBucket = [...ageCounter.entries()]
-    .sort((a, b) => {
-      if (b[1] !== a[1]) return b[1] - a[1];
-      return a[0].localeCompare(b[0]);
-    })[0]?.[0] ?? null;
-
-  const genderSkew = (() => {
-    const total = female + male;
-    if (total < 6) return benchmark.default_audience.gender_skew;
-    const femaleRatio = female / total;
-    const maleRatio = male / total;
-    if (femaleRatio >= 0.58) return '\u5973\u6027\u4e3b\u5bfc';
-    if (maleRatio >= 0.58) return '\u7537\u6027\u4e3b\u5bfc';
-    return '\u5747\u8861';
-  })();
-
-  return {
-    age_range: topAgeBucket ?? benchmark.default_audience.age_range,
-    gender_skew: genderSkew,
-    geo_focus: mergeDistinct(topItems(samples, 'city', 5), benchmark.default_audience.geo_focus, 5),
-    occupation_tags: mergeDistinct(topItems(samples, 'occupation', 5), benchmark.default_audience.occupation_tags, 5),
-    pain_points: mergeDistinct(topItems(samples, 'pain_point', 5), benchmark.default_audience.pain_points, 5),
-    aspirations: mergeDistinct(topItems(samples, 'aspiration', 5), benchmark.default_audience.aspirations, 5),
-  };
-}
-
-await startFixtureServer({
+await startThinProxy({
   serverId: 'audience-research',
   serverName: 'official-audience-research',
-  toolName: 'audience_research',
-  toolDescription: 'Build audience persona JSON from fixture fan snapshots plus optional uploaded CSV rows.',
-  inputSchema: {
-    account_id: z.string().min(1).describe('Target account identifier.'),
-    platform: z.enum(['xhs', 'douyin', 'wechat-mp']).optional()
-      .describe('Platform hint; inferred from account_id when omitted.'),
-    locale: z.string().optional().describe('Locale hint, default zh-CN.'),
-    follower_snapshot_csv: z.string().optional()
-      .describe('Optional CSV content. Headers supported: age,gender,city,occupation,pain_point,aspiration.'),
-    fan_samples: z.array(z.object({
-      age: z.union([z.number(), z.string()]).optional(),
-      gender: z.string().optional(),
-      city: z.string().optional(),
-      occupation: z.string().optional(),
-      pain_point: z.string().optional(),
-      aspiration: z.string().optional(),
-    })).optional().describe('Optional structured samples as an alternative to CSV upload.'),
-    sample_limit: z.number().int().min(20).max(400).optional()
-      .describe('Synthetic fan sample size for fixture estimation, default 90.'),
-  },
-  handler: ({
-    account_id,
-    platform,
-    locale = 'zh-CN',
-    follower_snapshot_csv,
-    fan_samples = [],
-    sample_limit = 90,
-  }) => {
-    const accountId = String(account_id ?? '').trim();
-    if (!accountId) throw new Error('account_id_required');
-
-    const normalizedPlatform = normalizePlatform(platform, accountId);
-    if (!Object.hasOwn(PLATFORM_BENCHMARKS, normalizedPlatform)) {
-      throw new Error(`unsupported_platform:${normalizedPlatform}`);
-    }
-
-    const benchmark = PLATFORM_BENCHMARKS[normalizedPlatform];
-    const normalizedLocale = String(locale ?? 'zh-CN').trim() || 'zh-CN';
-    const safeSampleLimit = clampInt(sample_limit, 20, 400, 90);
-
-    const csvSamples = parseCsv(follower_snapshot_csv, 300);
-    const directSamples = Array.isArray(fan_samples)
-      ? fan_samples.slice(0, 300).map((row) => normalizeSampleRow(row))
-      : [];
-
-    const syntheticSamples = generateSyntheticSamples({
-      accountId,
-      platform: normalizedPlatform,
-      count: safeSampleLimit,
-    });
-
-    const combinedSamples = [...syntheticSamples, ...directSamples, ...csvSamples];
-    const audience = inferAudience({
-      samples: combinedSamples,
-      benchmark,
-    });
-
-    return {
-      ...FIXTURE_META,
-      account_id: accountId,
-      platform: normalizedPlatform,
-      locale: normalizedLocale,
-      audience,
-      goal_schema_alignment: {
-        required_fields: ['age_range', 'gender_skew', 'geo_focus', 'occupation_tags', 'pain_points', 'aspirations'],
-        matched: true,
+  tools: [
+    {
+      name: 'audience_research',
+      description: 'Build audience persona JSON from fixture fan snapshots plus optional uploaded CSV rows.',
+      inputSchema: {
+        account_id: z.string().min(1),
+        platform: z.enum(['xhs', 'douyin', 'wechat-mp']).optional(),
+        locale: z.string().optional(),
+        follower_snapshot_csv: z.string().optional(),
+        fan_samples: z.array(z.object({
+          age: z.union([z.number(), z.string()]).optional(),
+          gender: z.string().optional(),
+          city: z.string().optional(),
+          occupation: z.string().optional(),
+          pain_point: z.string().optional(),
+          aspiration: z.string().optional(),
+        })).optional(),
+        sample_limit: z.number().int().min(20).max(400).optional(),
       },
-      source_mix: {
-        synthetic_fan_samples: syntheticSamples.length,
-        direct_fan_samples: directSamples.length,
-        csv_fan_samples: csvSamples.length,
-        public_benchmark_platform: normalizedPlatform,
-      },
-      notes: {
-        mvp_mode: 'semi_automatic_supported',
-        csv_supported: true,
-        confidence: (directSamples.length + csvSamples.length) >= 20 ? 'medium' : 'medium_low',
-      },
-      generated_at: new Date().toISOString(),
-    };
-  },
+    },
+  ],
 });