Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
/**
* Unified Search Service
*
* Fan-out search engine that queries all PortOS data sources in parallel.
* Uses Promise.allSettled for fault isolation — a failing adapter never
* blocks results from the other sources.
*
* Sources: Brain (inbox/people/projects/ideas/admin/memories/links/threads, served from
* the in-memory projections in brainSearchIndex.js), CoS Memory
* (BM25/hybrid), Apps, History, Health metrics
*/
import { getBrainProjections } from './brainSearchIndex.js';
import { searchBM25 } from './memoryBM25.js';
import { getMemories, ensureBackend, hybridSearchMemories } from './memoryBackend.js';
import { getAllApps } from './apps.js';
import { getHistory } from './history.js';
import { listProjects as listMusicVideoProjects } from './musicVideo/projects.js';
// =============================================================================
// SNIPPET HELPER
// =============================================================================
/**
* Extract a ~100-char window around the first keyword match in text.
* Prepends/appends '...' when the excerpt is not at the text boundary.
*/
function extractSnippet(text, query, maxLen = 100) {
if (!text) return '';
const lower = text.toLowerCase();
const idx = lower.indexOf(query.toLowerCase());
if (idx === -1) {
return text.substring(0, maxLen);
}
const start = Math.max(0, idx - 30);
const end = Math.min(text.length, start + maxLen);
const excerpt = text.substring(start, end);
const prefix = start > 0 ? '...' : '';
const suffix = end < text.length ? '...' : '';
return prefix + excerpt + suffix;
}
// =============================================================================
// SOURCE ADAPTERS
// =============================================================================
// Deep link into the Brain memory view for the id-addressed entity types.
const brainMemoryUrl = (type) => (record) => `/brain/memory?type=${type}&id=${record.id}`;
/**
* Brain sources, in the order their matches appear in the result list.
*
* `match` names the fields a query is tested against; they must stay a subset
* of the fields `brainSearchIndex` projects for that type, or the match
* silently stops working.
*/
const BRAIN_SOURCES = Object.freeze([
{
type: 'inbox',
resultType: 'inbox',
match: ['capturedText'],
title: (r) => (r.capturedText ?? '').substring(0, 60),
snippet: (r) => r.capturedText,
url: () => '/brain/inbox'
},
{
type: 'people',
resultType: 'person',
match: ['name', 'context'],
title: (r) => r.name,
snippet: (r) => r.context,
url: brainMemoryUrl('people')
},
{
type: 'projects',
resultType: 'project',
match: ['name', 'notes'],
title: (r) => r.name,
snippet: (r) => r.notes,
url: brainMemoryUrl('projects')
},
{
type: 'ideas',
resultType: 'idea',
match: ['title', 'oneLiner', 'notes'],
title: (r) => r.title,
snippet: (r) => r.oneLiner || r.notes,
url: brainMemoryUrl('ideas')
},
{
type: 'admin',
resultType: 'admin',
match: ['title', 'notes', 'nextAction'],
title: (r) => r.title,
snippet: (r) => r.notes || r.nextAction,
url: brainMemoryUrl('admin')
},
{
type: 'memories',
resultType: 'memory-entry',
match: ['title', 'content', 'mood'],
title: (r) => r.title,
snippet: (r) => r.content,
url: brainMemoryUrl('memories')
},
{
type: 'links',
resultType: 'link',
match: ['title', 'url', 'description'],
title: (r) => r.title || r.url,
snippet: (r) => r.description || r.url,
url: () => '/brain/links'
},
{
// A tracked open loop (#7664), not a message thread. `?thread=` opens the
// record's drawer on the Threads tab — the same deep-link the widget uses.
type: 'threads',
resultType: 'thread',
match: ['title', 'nextAction', 'waitingOn', 'notes'],
title: (r) => r.title,
snippet: (r) => r.nextAction || r.waitingOn || r.notes,
url: (r) => `/brain/threads?thread=${encodeURIComponent(r.id)}`
}
]);
/**
* Search the brain entity stores.
*
* Reads field projections from `brainSearchIndex` rather than
* `brainStorage.getAll()`, so a keystroke costs an in-memory scan instead of a
* full stat+read+parse of every record file across eight collection dirs
* (issue #3506). The index also removes the old `limit: 200` cap on inbox
* entries — that cap existed only to bound the per-query disk cost.
*/
async function searchBrain(query) {
const q = query.toLowerCase();
// allSettled per source: one unreadable store never blanks the others.
const settled = await Promise.allSettled(BRAIN_SOURCES.map((s) => getBrainProjections(s.type)));
const results = BRAIN_SOURCES.flatMap((source, i) => {
const outcome = settled[i];
if (outcome.status !== 'fulfilled') return [];
return (outcome.value ?? [])
.filter((r) => source.match.some((f) => typeof r[f] === 'string' && r[f].toLowerCase().includes(q)))
.map((r) => ({
id: r.id,
title: source.title(r),
snippet: extractSnippet(source.snippet(r), query),
url: source.url(r),
type: source.resultType
}));
}).slice(0, 8);
return { id: 'brain', label: 'Brain', icon: 'Brain', results };
}
async function searchMemory(query) {
const activeBackend = await ensureBackend();
// Postgres backend: use hybrid search (tsvector full-text, no embedding needed)
if (activeBackend === 'postgres') {
const searchResult = await hybridSearchMemories(query, null, { limit: 10 });
const results = (searchResult?.memories ?? [])
.map(mem => {
const summary = mem.summary ?? '';
return {
id: mem.id,
title: summary.substring(0, 60),
snippet: summary,
url: '/cos/memory',
type: 'memory'
};
})
.slice(0, 5);
return { id: 'memory', label: 'Memory', icon: 'Cpu', results };
}
// File backend: use BM25 index + getMemories for metadata
const [bm25Results, memoriesResult] = await Promise.allSettled([
searchBM25(query, { limit: 10, threshold: 0.1 }),
getMemories()
]);
if (bm25Results.status !== 'fulfilled' || memoriesResult.status !== 'fulfilled') {
return { id: 'memory', label: 'Memory', icon: 'Cpu', results: [] };
}
const hits = bm25Results.value ?? [];
const allMemories = memoriesResult.value?.memories ?? [];
const memoryMap = new Map(allMemories.map(m => [m.id, m]));
const results = hits
.map(hit => {
const mem = memoryMap.get(hit.id);
if (!mem) return null;
const summary = mem.summary ?? mem.content?.substring(0, 150) ?? '';
return {
id: mem.id,
title: summary.substring(0, 60),
snippet: summary,
url: '/cos/memory',
type: 'memory'
};
})
.filter(Boolean)
.slice(0, 5);
return { id: 'memory', label: 'Memory', icon: 'Cpu', results };
}
async function searchApps(query) {
const q = query.toLowerCase();
const apps = await getAllApps({ includeArchived: false });
const results = (apps ?? [])
.filter(a => a.name?.toLowerCase().includes(q) || a.description?.toLowerCase().includes(q))
.map(a => ({
id: a.id,
title: a.name,
snippet: extractSnippet(a.description, query),
url: '/apps',
type: 'app'
}))
.slice(0, 5);
return { id: 'apps', label: 'Apps', icon: 'Package', results };
}
async function searchHistory(query) {
const q = query.toLowerCase();
const { entries } = await getHistory({ limit: 500 });
const results = (entries ?? [])
.filter(e => e.targetName?.toLowerCase().includes(q) || e.action?.toLowerCase().includes(q))
.map(e => ({
id: e.id,
title: e.targetName || e.action,
snippet: extractSnippet((e.action ?? '') + ' ' + (e.targetName ?? ''), query),
url: '/devtools/history',
type: 'history'
}))
.slice(0, 5);
return { id: 'history', label: 'History', icon: 'History', results };
}
const HEALTH_METRICS = [
'step_count',
'heart_rate',
'sleep_analysis',
'hrv',
'blood_pressure',
'body_mass',
'respiratory_rate',
'blood_glucose',
'body_temperature'
];
const HEALTH_DISPLAY_NAMES = {
step_count: 'Steps',
heart_rate: 'Heart Rate',
sleep_analysis: 'Sleep',
hrv: 'HRV',
blood_pressure: 'Blood Pressure',
body_mass: 'Body Mass',
respiratory_rate: 'Respiratory Rate',
blood_glucose: 'Blood Glucose',
body_temperature: 'Body Temperature'
};
function searchHealth(query) {
const q = query.toLowerCase();
const results = HEALTH_METRICS
.filter(key => {
const displayName = HEALTH_DISPLAY_NAMES[key] ?? key;
return key.toLowerCase().includes(q) || displayName.toLowerCase().includes(q);
})
.map(key => {
const displayName = HEALTH_DISPLAY_NAMES[key] ?? key;
return {
id: key,
title: displayName,
snippet: `${displayName} health metric data`,
url: '/meatspace/health',
type: 'health-metric'
};
});
return { id: 'health', label: 'Health', icon: 'HeartPulse', results };
}
// A music-video concept is a structured object (prompt/style/...) or a legacy string.
// Pick a plain string for the snippet; never pass the object to extractSnippet.
function musicVideoSnippetText(p) {
const concept = p.concept;
if (typeof concept === 'string' && concept.trim()) return concept;
const candidates = [concept?.prompt, concept?.style, p.description, p.name];
return candidates.find(v => typeof v === 'string' && v.trim()) ?? '';
}
async function searchMusicVideo(query) {
const q = query.toLowerCase();
const projects = await listMusicVideoProjects({ includeDeleted: false }).catch(() => []);
const results = (projects ?? [])
.filter(p => p.name?.toLowerCase().includes(q))
.map(p => ({
id: p.id,
title: p.name,
snippet: extractSnippet(musicVideoSnippetText(p), query),
url: `/music-video/${p.id}`,
type: 'music-video'
}))
.slice(0, 5);
return { id: 'musicVideo', label: 'Music Video', icon: 'Video', results };
}
// =============================================================================
// FAN-OUT ENGINE
// =============================================================================
const ADAPTERS = ['brain', 'memory', 'apps', 'history', 'health', 'musicVideo'];
/**
* Fan out a keyword query to all PortOS data sources in parallel.
* Returns an array of non-empty source result objects.
*/
export async function fanOutSearch(query) {
console.log(`🔍 Search fan-out for "${query}" across ${ADAPTERS.length} sources`);
const [brainResult, memoryResult, appsResult, historyResult, healthResult, musicVideoResult] =
await Promise.allSettled([
searchBrain(query),
searchMemory(query),
searchApps(query),
searchHistory(query),
Promise.resolve(searchHealth(query)),
searchMusicVideo(query)
]);
const FALLBACKS = [
{ id: 'brain', label: 'Brain', icon: 'Brain', results: [] },
{ id: 'memory', label: 'Memory', icon: 'Cpu', results: [] },
{ id: 'apps', label: 'Apps', icon: 'Package', results: [] },
{ id: 'history', label: 'History', icon: 'History', results: [] },
{ id: 'health', label: 'Health', icon: 'HeartPulse', results: [] },
{ id: 'musicVideo', label: 'Music Video', icon: 'Video', results: [] }
];
const settled = [brainResult, memoryResult, appsResult, historyResult, healthResult, musicVideoResult];
const sources = settled.map((r, i) => r.status === 'fulfilled' ? r.value : FALLBACKS[i]);
const nonEmpty = sources.filter(s => s.results.length > 0);
const totalResults = nonEmpty.reduce((sum, s) => sum + s.results.length, 0);
console.log(`✅ Search complete for "${query}": ${totalResults} results across ${nonEmpty.length} sources`);
return nonEmpty;
}