feat(p7): ai-studio demo + smoke-ai; P7 verification green
Task 7.6: apps/ai-studio (Vite, port 5177) — ingest a message, run CLASSIFY/ SUMMARIZE/EXTRACT, see proposal cards with confidence, evidence citations, abstention, model/cost + a budget meter, and accept/reject; built on the ai-web SDK via AiProvider. scripts/smoke-ai.mjs proves the P7 guarantees end to end: auto-CLASSIFY → AI flag → routing REVIEW/DENY never ALLOW (KG-05), summary cites evidence, EXTRACT stays DISCUSSION (Scenario 7), replay-once, accept. P7 verification: 79 tests green (3x stable), pnpm -r build all packages+demos, boundary passes and fails on ai-web->service, smoke-ai PASS, prior smokes (realtime/inbox/support/adapter/route) PASS. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
@@ -0,0 +1,97 @@
|
||||
// P7 AI smoke: proposal-only. Auto-CLASSIFY → AI flag → routing REVIEW/DENY never
|
||||
// ALLOW (KG-05); SUMMARIZE cites evidence; EXTRACT stays DISCUSSION (Scenario 7);
|
||||
// fail-closed; budget degrade (KG-12); replay-once; accept. No real network.
|
||||
import 'dotenv/config';
|
||||
import crypto from 'node:crypto';
|
||||
import jwt from 'jsonwebtoken';
|
||||
|
||||
const SERVICE = process.env.SMOKE_URL ?? 'http://localhost:3200';
|
||||
const APP_ID = 'portal-demo';
|
||||
const APP_SECRET = JSON.parse(process.env.APP_SECRETS ?? '{"portal-demo":"dev-secret"}')[APP_ID];
|
||||
const ADAPTER_SECRET = (() => {
|
||||
try { return JSON.parse(process.env.ADAPTER_SECRETS ?? '{}').WEBHOOK ?? 'dev-adapter-secret'; } catch { return 'dev-adapter-secret'; }
|
||||
})();
|
||||
|
||||
const token = jwt.sign({ sub: 'ai-studio', appId: APP_ID, orgId: `org_${APP_ID}` }, APP_SECRET, { algorithm: 'HS256', expiresIn: '1h' });
|
||||
const sign = (body) => 'sha256=' + crypto.createHmac('sha256', ADAPTER_SECRET).update(body).digest('hex');
|
||||
const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
|
||||
const assert = (c, m) => { if (!c) { console.error('✗', m); process.exit(1); } console.log('✓', m); };
|
||||
|
||||
async function req(path, method = 'GET', body) {
|
||||
const r = await fetch(`${SERVICE}${path}`, {
|
||||
method,
|
||||
headers: { 'content-type': 'application/json', authorization: `Bearer ${token}` },
|
||||
body: body ? JSON.stringify(body) : undefined,
|
||||
});
|
||||
if (!r.ok) throw new Error(`${method} ${path} ${r.status}: ${await r.text()}`);
|
||||
return r.json();
|
||||
}
|
||||
async function pollUntil(fn, ms = 8000) {
|
||||
const end = Date.now() + ms;
|
||||
while (Date.now() < end) { const v = await fn(); if (v) return v; await sleep(300); }
|
||||
return null;
|
||||
}
|
||||
|
||||
async function ingest(text) {
|
||||
const payload = { eventId: `ai-sm-${crypto.randomUUID()}`, from: 'sim@member', text, threadRef: `sm-${Date.now()}` };
|
||||
const body = JSON.stringify(payload);
|
||||
const wh = await fetch(`${SERVICE}/v1/adapters/WEBHOOK/webhook`, {
|
||||
method: 'POST',
|
||||
headers: { 'content-type': 'application/json', 'x-iios-signature': sign(body) },
|
||||
body,
|
||||
});
|
||||
if (wh.status !== 202) throw new Error(`webhook ${wh.status}`);
|
||||
const e = await pollUntil(async () => {
|
||||
const list = await req('/v1/adapters/inbound');
|
||||
const hit = list.find((x) => x.externalEventId === payload.eventId);
|
||||
return hit && hit.interactionId ? hit : null;
|
||||
});
|
||||
if (!e) throw new Error('interaction never normalized');
|
||||
return e.interactionId;
|
||||
}
|
||||
|
||||
// ── KG-05: inbound "party at 9 PM" is auto-classified; AI flag routes to REVIEW/DENY, never ALLOW ──
|
||||
const interactionId = await ingest('There is a party at 9 PM tonight!');
|
||||
|
||||
// The AiProjector auto-runs CLASSIFY on interaction.normalized — wait for the AI flag.
|
||||
const flagged = await pollUntil(async () => {
|
||||
const arts = await req(`/v1/ai/artifacts?interactionId=${interactionId}`);
|
||||
return arts.find((a) => a.artifactType === 'CLASSIFICATION') ?? null;
|
||||
}, 10000);
|
||||
assert(flagged, 'inbound message auto-classified by the AiProjector (proposal)');
|
||||
assert(flagged.claims.some((c) => c.claimType === 'MODERATION_FLAG'), 'CLASSIFICATION proposes a MODERATION_FLAG claim');
|
||||
|
||||
// Route it to an unrestricted auto-ALLOW binding + a CHILD binding.
|
||||
const adults = await req('/v1/routes/bindings', 'POST', { originChannelType: 'WEBHOOK', originRef: 'webhook', destinationChannelType: 'PORTAL', destinationRef: 'adults', requiresReview: false, enabled: true });
|
||||
const children = await req('/v1/routes/bindings', 'POST', { originChannelType: 'WEBHOOK', originRef: 'webhook', destinationChannelType: 'PORTAL', destinationRef: 'children', restrictionProfile: 'CHILD', requiresReview: false, enabled: true });
|
||||
const before = (await req('/v1/adapters/outbound')).length;
|
||||
const { decisions } = await req('/v1/routes/simulate', 'POST', { interactionId, originChannelType: 'WEBHOOK', originRef: 'webhook' });
|
||||
const state = (id) => decisions.find((d) => d.routeBindingId === id)?.decisionState;
|
||||
assert(state(adults.id) === 'REVIEW', 'AI flag forces adults → REVIEW (never ALLOW) — KG-05');
|
||||
assert(state(children.id) === 'DENY', 'AI flag on CHILD → DENY (deny-by-default)');
|
||||
assert((await req('/v1/adapters/outbound')).length === before, 'AI proposal + simulate sent nothing');
|
||||
|
||||
// ── SUMMARIZE cites evidence ──
|
||||
const sum = await req('/v1/ai/jobs', 'POST', { interactionId, jobType: 'SUMMARIZE' });
|
||||
assert(sum.artifact.artifactType === 'SUMMARY' && String((sum.artifact.contentRef ?? {}).text).includes('[ai]'), 'SUMMARIZE → [ai] summary artifact');
|
||||
assert((sum.artifact.evidence ?? []).some((e) => e.sourceId === interactionId), 'summary cites the source interaction');
|
||||
|
||||
// ── EXTRACT stays DISCUSSION (Scenario 7) ──
|
||||
const ext = await req('/v1/ai/jobs', 'POST', { interactionId, jobType: 'EXTRACT' });
|
||||
const eventClaim = (ext.artifact.claims ?? []).find((c) => c.claimType === 'EVENT');
|
||||
assert(eventClaim && eventClaim.claimJson.certainty === 'DISCUSSION', 'EXTRACT event certainty=DISCUSSION (never CONFIRMED) — Scenario 7');
|
||||
|
||||
// ── replay a job twice → one artifact ──
|
||||
const safe = await ingest('The committee met today. Budget discussed.');
|
||||
await req('/v1/ai/jobs', 'POST', { interactionId: safe, jobType: 'SUMMARIZE' });
|
||||
const a1 = await req(`/v1/ai/artifacts?interactionId=${safe}`);
|
||||
await req('/v1/ai/jobs', 'POST', { interactionId: safe, jobType: 'SUMMARIZE' });
|
||||
const a2 = await req(`/v1/ai/artifacts?interactionId=${safe}`);
|
||||
assert(a1.filter((a) => a.artifactType === 'SUMMARY').length === 1 && a2.filter((a) => a.artifactType === 'SUMMARY').length === 1, 'replaying a job is idempotent (one artifact)');
|
||||
|
||||
// ── accept records human feedback ──
|
||||
const accepted = await req(`/v1/ai/artifacts/${sum.artifact.id}/accept`, 'POST');
|
||||
assert(accepted.status === 'ACCEPTED', 'accept → status ACCEPTED (human feedback, not a send)');
|
||||
|
||||
console.log('\nP7 AI smoke: PASS');
|
||||
process.exit(0);
|
||||
Reference in New Issue
Block a user