chore(gui): mission work logs

chore(gui): mission work logs — cordis design finalization, tool-card wire archive, incident records

chore: missions

chore: missions

chore(gui): mission ledger — batch-2 answers, parallel dispatch state, jsdom coverage re-scope

chore(gui): ledger — night-mode standing orders (self-commit small, no push, 5-min refresh)

chore(gui): ledger — jsdom batches 2-4 landed (233 green), coverage probe next

chore(gui): ledger — web-ui coverage probe 65%, four-tier fill plan approved

chore(gui): ledger — cordis-impl B1 state after third API drop, decisions on file

chore(gui): ledger — 01:32 patrol snapshot (jsdom tier-1 landed, coverage-fixer probed)

chore(gui): ledger — 01:37 patrol (peer src trio landed, coverage-fixer still silent)

chore(gui): ledger — 01:42 patrol (jsdom tier-2 landed, peer committed x2, coverage-fixer 2nd probe)

chore(gui): ledger — coverage diagnosis complete (6-file gap list), web-ui at 91.4%

chore(gui): ledger — 01:46 patrol (B1 done, jsdom tier-3 landed, coverage fix batch running)

chore(gui): ledger — 01:51 patrol (jsdom tails x2 landed, B2 underway)

chore(gui): ledger — 01:56 patrol (gateway.ts 337 lines, checkpoint T-7min)

chore(gui): ledger — hold/pending split ruling, coverage-fixer externalize-or-restart ultimatum

chore(gui): ledger — 02:01 patrol (B2 done, jsdom final arms, coverage ultimatum pending)

chore(gui): ledger — 02:03 checkpoint executed (fixer2 respawn, four lanes released, three owners cold-started)

chore(gui): ledger — all six lanes acked, type isolation first live proof (client closure clean)

chore(gui): ledger — 02:08 patrol (all seven lanes active, wire carrier assembled)

docs(gui): respond-design task checkpoint — apiproxy wire-layer recon done

chore(gui): ledger — P0-2 contributor AGENTS.md landed (dd28a5019)

docs(gui): respond-design checkpoint 2 — host-side recon (stub respond, frame types, approval seam, ACP answerer precedent)

docs(gui): OOP debt inventory — seven territories, 2 real debts (createApiProxy, createFixtureApi), rest ruled keep-as-is

docs(gui): disambiguation note on the archived i18n design task

chore(gui): ledger — 02:12 patrol (exclude removed, mixed-knife incident under reconciliation)

chore(gui): ledger — 02:15 wave (jsdom mission closed, OOP audit done, B3 isolation proof, mixed-knife resolved)

chore(gui): ledger — 02:17 patrol (attribution reversal filed, arch-session probed)

chore(gui): ledger — 02:22 patrol (B4 done, B5+B6 merged batch, arch-session deadline set)

docs(gui): respond-design checkpoint 3 — client-side recon (pending map, PendingCard onRespond stub, AbstractApiClient.respond ready, bootHost missing approval mounts)

docs(gui): peer carrier territory review — 2 fixes (SSE cancel leak, route-reservation guard), 1 ruling ask (RPC-log visibility), compliance ledger

chore(gui): ledger — 02:27 (arch-shell respawn, territory review verdicts routed, ask-deny finding flagged)

docs(gui): P1-5 respond design page complete — pending registry, wire answerer, client state machine, first-wins arbitration

chore(gui): ledger — 02:31 patrol (respond design complete, B5 wire smoke green, shell knife 1 underway)

docs(gui): respond design — add §0 status warning (web host ask defaults to deny), mount-behavior delta, no-timeout ruling with Config discipline

chore(gui): ledger — 02:42 patrol (respond line closed pending review, B7 last piece underway)

docs(gui): respond design contract review — direction pass, 2 doc fixes (settle-order contradiction, answering-state race), A/B/C compliance ledger

docs(gui): respond design — contract review fixes (R1 verify-before-delete arbitration, R2 answering+resolved-frame transition, ask dual-source wording, rejected-is-ok-value note)

chore(gui): ledger — 02:46 patrol (respond line final, two user decisions distilled, arch-shell deadline)

docs(gui): respond review addendum — contract-gap ruling: approve plan A (ApprovalRequest.id), wire unchanged, drop plan B backscan

chore(gui): ledger — cordis B7 summit: real-browser 10/10 green, user acceptance criterion proven

docs(gui): respond design — contract gap #4 approved as plan A (ApprovalRequest.id), backscan fallback retired, blade 0 prepended

docs(gui): respond design — final polish (owner-approval vs user-go-ahead wording, implementation handoff notes)

chore(gui): ledger — 02:51 (shell-exec third respawn with operational script, respond line 4-knife final)

chore(gui): ledger — 02:53 wave (respond five-knife true final, B7 closed 12/12, client.ts green)

chore(gui): ledger — 02:56 patrol (cordis closeout bounced pending R1/R2/N1, shell-exec first sign of life)

chore(gui): ledger — 03:01 patrol (R1/R2/N1 remediation in flight across four files)

chore(gui): ledger — cordis line officially closed and archived, verified on disk (24 knives, 12/12, reviews closed)

chore(gui): ledger — 03:06 patrol (shell-exec final window, lowered first-knife bar)

chore(gui): ledger — 03:11 patrol (shell line iced-broken: three registries on disk)

chore(gui): ledger — 03:15 patrol (quiet window, both active lanes within threshold)

chore(gui): ledger — 03:20 patrol (shell five files up, api-proxy plan reported)

chore(gui): ledger — 03:25 patrol (shell migration in flight with history-preserving moves, webserver green)

chore(gui): ledger — 03:30 patrol (shell knife-1 in verification, api-proxy patching)

chore(gui): ledger — shell knife 1 accepted (694cecc53), knife 2 released

chore(gui): ledger — 03:40 patrol (knife 2 pre-move stage, cold-list spec appears)

chore(gui): ledger — 03:45 patrol (rpclog moves staged, api-proxy two specs in flight)

chore(gui): ledger — 03:54 patrol (knife-2 code done, coverage full-run final check)

docs(gui): coverage-fixer task ledger — fixer2 takeover, per-file fix log, isolated reportsDirectory pitfall

chore(gui): ledger — coverage lane closed and accepted (a19f069a5), the PR #443 CI fix knife

chore(gui): ledger — 04:04 patrol (knife-2 calibration, sole active lane)

chore(gui): ledger — shell knife 2 accepted (f8fb77b95), knife 3 released as final night task

chore(gui): ledger — 04:19 patrol (knife-3 past half: callback chain through, ToolCallDetail up)

chore(gui): ledger — night closeout summary: seven lanes closed, wake-up decision sheet

chore: missions

chore: missions

chore: missions

chore(gui): mission-local browser/probe verify scripts under missions/scripts/

The six acceptance/probe scripts move here as mission-side working
material (headers and relative imports adjusted for the new location):
carrier-errors, rpclog-panel, session, session-real,
webserver-backpressure, webserver-hardening.

chore(gui): verify-relocate mission log

chore(gui): verify-relocate mission log — R1 guard addendum

chore(gui): gates-continue mission log — CI-equivalent sequence all green

chore(gui): VS Code 扩展体系双边设计调研报告

chore(gui): 调研追加 4.5 节——git 扩展数据面与 scope 绑定

docs(gui): web plugin system RFC — walkthrough + design notes

docs(gui): RFC — restore existing SSE/POST as the v1 transport; envelope rides on it (D16)

docs(gui): RFC — envelope demoted to chan-dispatch, scope out of envelope, rpc-log cut, peer deferred, scope tree is native cordis (D17-D21)

docs(gui): RFC — hooks re-derived from component needs: useWatch/useAction only, useService removed; sessionHub cut, projections user-space, router rename, loader-only root (D22-D25)

docs(gui): RFC — drop stale fork vocabulary (vendored cordis has Fiber only; scope = mintScope pattern), hook idempotence contract (D26-D27)

docs(gui): RFC — session precision seam: plugins read scope key (host paradigm), React gets it from tree position via SlotOutlet (D28)

docs(gui): RFC — domain hooks owned by plugins over framework primitives; useConversation paradigm carried over (D29)

docs(gui): RFC — ctx services are the inter-plugin API (cordis proper); declarations are wire-only; get(id) returns scoped ctx (D30)

docs(gui): RFC — full ctx.conversation walkthrough: root-singleton scope-sensitive service, caller-ctx scope key, get(key) as scoped ctx (D31)

docs(gui): RFC — no client-side agents collection: session state machine already expresses the duality; agent resolution stays host authority (D32)

docs(gui): RFC — v1 stays session-precision, no agent-level isolation; incarnation/agent-axis designs archived in ledger (D33)

docs(gui): RFC walkthrough — full rewrite to final state (D16-D33 consolidated), end-to-end chain restored

docs(gui): RFC — apiproxy demoted to generic channel routing; domain RPCs dissolve into owner plugins (D34)

docs(gui): RFC — TS-interface-first wire contract (zod internal), conversation owns the dialogue frame with pluggable views (D35)

docs(gui): RFC — page skeleton (sidebar+conversation), projects as plugin not service, nested slots via owner registries (D36)

docs(gui): RFC — SlotMap declaration-merging slot model: single register API, inject-as-ownership, FC-typed registration, typed outlets (D37)

docs(gui): RFC — slot props whitelist: identity, display params, materialized snapshot slices, stable UI callbacks (D38)

docs(gui): RFC — end-to-end data flow: three transforms, equality protocol table, immer placement; i18n/theme kept standard (D39-D40)

docs(gui): RFC — full external-injection model: props carry values + stable injected hooks; shared/client/react example rewritten (D41-D43)

docs(gui): RFC final trio — modules.md (agent implementation spec), architecture.md (human walkthrough), plugins.md (business plugin inventory)

docs(gui): RFC — props three-source merge (scope-standard useSession auto-injected); keyed key vs list id disambiguated (D44)

docs(gui): RFC — inject comment says what it is (the React-facing props bundle); SessionHandle rename; snapshot-production story unified on buildSnapshot

docs(gui): RFC architecture — full React component tree walkthrough: props three sources, slot vs plain children, hook taxonomy per node

docs(gui): RFC — module map finalized (ui-slots/web-react/connection/runtime/ui-*/web); slots onChange replaced by cordis events; toolcall dimension; detail sidebar default-collapsed with toolName-keyed routing

docs(gui): RFC plugins — openDetail relay chain: toolcard calls chat-view injected action, chat-view relays to conversation sidebar

chore(gui): progress ledger — full archive rewrite: RFC outcome digest, open gaps, dispatch plan, cold-start entry

docs(gui): RFC grill pass 1 — SlotScope axis (root/session) on declares, Gate dependency inversion, inject handle by scope, W5 acceptance list, gantt relay chain fixed

docs(gui): figma analysis — sidebar/projects/sessions 区域交互视觉理解报告

docs(gui): figma 解析报告 — details 面板/多视图 tabs/未来功能区盘点 + slot 需求清单

docs(gui): figma 对话主区解析报告 — 消息流/tool calls 变体/审批接管输入框/Header tabs/视觉 token

docs(gui): plugins.md rewritten from figma analysis — three-column layout, full slot reservation table, selection channel, composer-takeover approvals, phased scope

docs(gui): layout dynamics ruled (drag+collapse both rails, details yields first, composer swap-panel, same-component transition); toolviews promoted to named scope-aware registry

docs(gui): P-I scope locked (details minimal, dual theme, chat-view, custom toolview sample); teammate dispatch plan — 6 owners by package, dependency-driven waves, contract arbitration

docs(gui): P-I api-contracts (full inter-package API spec) + dispatch plan (T0 skeleton knife, 7-dev roster, task briefs, milestones)

docs(gui): api-contracts v2 — scope tree in P-I, bundle loader + per-plugin CSS isolation in P-I, agent-scoped toolviews live, zustand engine, renames (SessionProvider/ObservableSnapshot/SessionBinding), router owns all shell view-state

docs(gui): services roster + progressive loading (no blocking loadAll), SlotsService as real cordis Service, renderSlot/renderSuspenseSlot duo, ui-traj teammate

docs(gui): loading-chain gaps ruled — dev=rebundle no HMR, ui-primitives package, externals on globals (no import map), host injects __DSH_BOOT__ into HTML (zero round-trip)

docs(gui): api-contracts v3 + dispatch v2 final — 12 packages, services merged in, progressive loader, global externals, __DSH_BOOT__ injection, 8-dev roster with convo split and ui-traj

docs(gui): v3 amendments — router renamed ctx.layout, ui-trajectory has no service (pure consumer sample), wait-for-settled loading (no Suspense in P-I, ledger 6b)

chore(gui): progress — pre-compact final state: v3 revision chain, 8-dev roster, T0 procedure, doc authority order

docs(gui): authority banners — modules/architecture get v3 term-mapping headers, walkthrough marked as archived process doc

docs(gui): cssdesign token set is THE theme source (--dsw-* variables, data-ds-dark-theme switch); recorded in contracts + progress

docs(gui): architecture.md full v3 rewrite — loading chain, 12-package map, service roster, slot/inject/toolviews, data flow, component tree, perf model, all current

docs(gui): contracts — UI plugins are dual-entry host plugins (node half serves client asset via ctx.webPlugins; __DSH_BOOT__ derives from it; client-closure gate back in scope)

docs(gui): contracts — closure-factory bundles with DI require (no globals), package.json dshWeb declarative discovery (no serve ritual), create-then-send empty state with project picker, ancestry() for breadcrumb, unload stubbed until HMR, props.renderSlot confirmed

docs(gui): dshClient declaration (inject/platform/immediately, exports./client), closure-DI require loading — synced across contracts/dispatch/modules/architecture/walkthrough

chore(gui): progress — record final loading-chain rulings (dshClient declaration, closure-DI require, startSession) before compact

docs(gui): architecture.md — developer-facing whole-web architecture on master baseline 6b16a67cb: what exists, what is new, no process narrative

docs(gui): architecture.md — self-contained whole-web architecture: absorbs still-valid substance from the branch RFCs (host layering, four-quadrant RPC, object layer, testing tiers) under the new plugin system as the override

chore(gui): progress — final pre-compact snapshot: contracts digest, apiproxy purity ruling, T0 procedure with first-action list

docs(gui): api-contracts v3 §3.1 — apiproxy purity principle with three-way existing-code verdicts

docs(gui): api-contracts v3 — immediately reinterpreted as static-infra group (8-package dshClient scope, boot manifest reconciliation)

docs(gui): api-contracts v3 — immediately corrected to early-load dynamic group (prod shell must not rebundle); loader shell-held; bundles register their export surface into module table

docs(gui): architecture — align with immediately=early-load dynamic group ruling; loader shell-held; module-table registration of loaded bundles

docs(gui): T0 checklist — 12-package skeleton table, 4-cut sequence, mv/attic/rewire rules (pre-drafted, awaiting go)

docs(gui): t0-checklist — pin figma-flows findings (missing font-family base vars, three alias vars behind upstream)

docs(gui): dispatch v2.1 — drop cordis-web salvage wording, two-wave staffing, loader/immediately boundary updates

docs(gui): progress + t0-checklist ledger — T0 landed, staffing status, execution accounting

docs(gui): api-contracts v3 — arbitration round 1: renderBody deps, RootBindingProvider, flush default sync, prune current, loader subpath, config-source P-I bar

docs(gui): v3 §3.2 connection 导出清单附录(rt-core 对账)+ rt-core 实现计划档案

docs(gui): progress — T1 milestone, arbitration round 1 ledger, fw-react timeout escalation

docs(gui): progress rolling update — per-line battlefield state at 00:2x, mailbox-vs-contract lesson, small-batch discipline reinforced

docs(gui): fw-react notes — v3 §2 complete, seven knives, T1/T2 follow-ups

docs(fw-slots): archive — four packages landed, open tails logged

docs(gui): progress — framework layer complete (web-react five, fw-slots four packages), T2 gated on rt-core runtime knife only

docs: api-contracts

docs: style-spec

docs

docs(gui): tsconfig convergence ruling — no host.json, root resumes host-aggregate duty, typecheck = root + client aggregates

docs(gui): missions 根三份 07-18 世代档案加「已被取代」头注——指向 web-plugin-rfc 现行权威并注明新旧对应

missions

docs(gui): progress rewritten for post-closeout state — wave ledger, architecture finale, teammate roster with handover notes, pending-user-command queue
This commit is contained in:
imccyu
2026-07-20 18:50:24 +08:00
committed by _Kerman
parent 402f8c5f5a
commit 0681ac47de
165 changed files with 17001 additions and 0 deletions

View File

@@ -0,0 +1,113 @@
// Carrier error-channel regression probes (audit batch: A1/A2/A4/A9 + R2 half).
// Runs the isomorphic path (InProcessApiClient over toFetchHandler) — no server needed.
// Run: node --experimental-strip-types missions/scripts/verify-carrier-errors.mjs (or via tsx)
import { toFetchHandler } from '../../packages/host/apiproxy/src/fetch/handler.ts'
import { InProcessApiClient } from '../../packages/host/apiproxy/src/fetch/client.ts'
import { RpcId } from '../../packages/host/apiproxy/src/api/rpc.ts'
import { serverResponseSchema } from '../../packages/host/apiproxy/src/api/rpc.schema.ts'
let failures = 0
const report = (n, p, d = '') => { failures += p ? 0 : 1; console.log(`${p ? 'PASS' : 'FAIL'} ${n}${d ? ' — ' + d : ''}`) }
const okList = { rpcId: RpcId('x'), result: { ok: true, value: { items: [] } } }
/** Minimal ApiProxy stub; per-test cases override single methods. */
function makeApi(overrides = {}) {
return {
sessions: {
list: async (r) => ({ ...okList, rpcId: r.rpcId }),
create: async (r) => ({ rpcId: r.rpcId, result: { ok: true, value: { sessionId: 's1' } } }),
history: async (r) => ({ rpcId: r.rpcId, result: { ok: true, value: { events: [], hasMore: false } } }),
prompt: async (r) => ({ rpcId: r.rpcId, result: { ok: true, value: { accepted: true } } }),
cancel: async (r) => ({ rpcId: r.rpcId, result: { ok: true, value: { accepted: true } } }),
...overrides.sessions,
},
host: {
describe: async (r) => ({ rpcId: r.rpcId, result: { ok: true, value: { version: '0', cwd: '/', attachedSessions: 0 } } }),
...overrides.host,
},
events: {
mux: overrides.mux ?? async function* () {},
host: overrides.hostStream ?? async function* () {},
},
respond: async () => ({ accepted: false, reason: 'not-pending' }),
}
}
// ---- A1: mid-stream impl throw → one stream/error frame on the wire, then clean close ----
{
const api = makeApi({
hostStream: async function* () {
yield { rpcId: RpcId('f1'), payload: { type: 'host/session-status', sessionId: 's1', running: true } }
throw new Error('impl exploded mid-stream')
},
})
const client = new InProcessApiClient(toFetchHandler(api))
const seen = []
for await (const frame of client.events.host({}, new AbortController().signal)) seen.push(frame.payload)
report('A1 流中 impl throw → stream/error 帧真到达 client', seen.some(f => f.type === 'stream/error' && f.error.code === 'internal' && /impl exploded/.test(f.error.message)), JSON.stringify(seen.map(f => f.type)))
report('A1b stream/error 后流正常收尾(迭代自然结束不 throw', true)
}
// ---- A2: S→C frame validation — a malformed frame is dropped, the stream survives ----
{
const api = makeApi({
hostStream: async function* () {
yield { rpcId: RpcId('bad'), payload: { type: 'host/session-status', sessionId: 's1' } } // missing `running`
yield { rpcId: RpcId('good'), payload: { type: 'host/session-status', sessionId: 's1', running: false } }
},
})
const client = new InProcessApiClient(toFetchHandler(api))
const seen = []
for await (const frame of client.events.host({}, new AbortController().signal)) seen.push(frame)
report('A2 坏帧被丢弃且不杀流(后续好帧照常到达)', seen.length === 1 && seen[0].payload.running === false, `seen=${seen.length}`)
}
// ---- A2: S→C unary value validation — a wrong-shaped ok value throws at the client boundary ----
{
const api = makeApi({ sessions: { list: async (r) => ({ rpcId: r.rpcId, result: { ok: true, value: { items: 'not-an-array' } } }) } })
const client = new InProcessApiClient(toFetchHandler(api))
const threw = await client.sessions.list({}).then(() => false, () => true)
report('A2b unary ok value 过 Value schema坏形状在 client 边界抛出)', threw)
}
// ---- A4: envelope parse failure backfills a salvageable rpcId; otherwise the sentinel — and the response parses as a valid ServerResponse ----
{
const handler = toFetchHandler(makeApi())
const post = (body) => handler.fetch('http://dsh.internal/api/session.list', {
method: 'POST', headers: { 'content-type': 'application/json' }, body: JSON.stringify(body),
})
const salvaged = await (await post({ rpcId: 'my-id', method: 5 })).json()
report('A4 信封烂但 rpcId 可捞 → 回填原值', serverResponseSchema.safeParse(salvaged).success && salvaged.rpcId === 'my-id', JSON.stringify(salvaged.rpcId))
const sentinel = await (await post({ nothing: true })).json()
report('A4b rpcId 不可捞 → invalid-request 哨兵,且过 serverResponseSchema', serverResponseSchema.safeParse(sentinel).success && sentinel.rpcId === 'invalid-request', JSON.stringify(sentinel.rpcId))
}
// ---- A10: external signal aborts an in-flight unary ----
{
const api = makeApi({ sessions: { list: () => new Promise(() => {}) } })
const client = new InProcessApiClient(toFetchHandler(api))
const ctl = new AbortController()
const call = client.sessions.list({}, ctl.signal).then(() => 'resolved', (e) => String(e))
ctl.abort(new Error('user cancelled'))
const outcome = await call
report('A10 unary 外部 signal 可取消在途请求', outcome !== 'resolved', outcome.slice(0, 60))
}
// ---- onOpen: stream-established signal fires before any frame is delivered ----
{
const api = makeApi({
hostStream: async function* () {
yield { rpcId: RpcId('f'), payload: { type: 'host/session-removed', sessionId: 's1' } }
},
})
const client = new InProcessApiClient(toFetchHandler(api))
const order = []
const iter = client.events.host({}, new AbortController().signal, () => order.push('open'))[Symbol.asyncIterator]()
await iter.next()
order.push('frame')
report('C2 信号onOpen 先于首帧交付', order.join(',') === 'open,frame', order.join(','))
await iter.return?.()
}
console.log(failures === 0 ? 'ALL PASS' : `${failures} FAILURE(S)`)
process.exit(failures === 0 ? 0 : 1)

View File

@@ -0,0 +1,99 @@
// RPC panel browser acceptance (fixture mode); step tags §D-1..§D-6 match the report labels.
// Prereqs: dsh web running on 3080, apps/web/dist freshly built, playwright chromium installed.
// Run: node missions/scripts/verify-rpclog-panel.mjs (not part of any gate system)
import { chromium } from 'playwright'
const BASE = process.env.DSH_WEB_URL ?? 'http://127.0.0.1:3080'
let failures = 0
function report(name, pass, detail = '') {
failures += pass ? 0 : 1
console.log(`${pass ? 'PASS' : 'FAIL'} ${name}${detail ? `${detail}` : ''}`)
}
const browser = await chromium.launch()
try {
const page = await browser.newPage()
await page.goto(`${BASE}/?fixture`, { waitUntil: 'load' })
// §D-1 page shell + rpclog rail button present, unread badge > 0 (boot auto-ping + subscribed frames); shell presence = the list sidebar.
await page.waitForSelector('aside')
const railBtn = page.locator('nav button[title="RPC 日志"]')
await railBtn.waitFor({ state: 'visible' })
await page.waitForFunction(() => {
const el = document.querySelector('nav button[title="RPC 日志"] span[class*="unread"]')
return el !== null && /\d/.test(el.textContent ?? '')
}, undefined, { timeout: 5000 })
report('§D-1 角标存在且未读数 > 0', true)
// §D-2 activate the rpclog bar: the ledger page fills the panel area, kinds cover three quadrants (direction symbols ↑ ↓ ⇟, up/down spatial metaphor)
await railBtn.click()
const list = page.locator('section:has(header)').locator('div[class*="list"]')
await list.waitFor({ state: 'visible' })
const rowTexts = await list.locator('button[class*="rowLine"]').allTextContents()
const joined = rowTexts.join('\n')
const hasThree = joined.includes('↑') && joined.includes('↓') && joined.includes('⇟')
report('§D-2 展开见台账,三象限方向符齐', hasThree, `rows=${rowTexts.length}`)
const unreadAfterOpen = await page.locator('span[class*="unread"]').count()
report('§D-2 展开后未读徽标消失', unreadAfterOpen === 0)
// §D-3 click ping: adds one client-request/server-response pair (host.describe)
const rowsBefore = await list.locator('button[class*="rowLine"]').count()
await page.locator('button', { hasText: 'ping' }).click()
await page.waitForFunction(
(n) => document.querySelectorAll('button[class*="rowLine"]').length >= n + 2,
rowsBefore, { timeout: 3000 },
)
const lastTwo = (await list.locator('button[class*="rowLine"]').allTextContents()).slice(-2)
const pingPair = lastTwo[0]?.includes('host.describe') && lastTwo[0]?.includes('↑')
&& lastTwo[1]?.includes('host.describe') && lastTwo[1]?.includes('↓')
report('§D-3 ping 新增一对 describe 往返', Boolean(pingPair), lastTwo.map((t) => t.slice(0, 30)).join(' | '))
// §D-3b hover pair highlight: hovering the last row (server-response) lights its client-request row too
const rows = list.locator('div[class*="row"]:not([class*="rowLine"])')
await list.locator('button[class*="rowLine"]').last().hover()
await page.waitForTimeout(100)
const pairedCount = await list.locator('div[class*="rowPaired"]').count()
report('§D-3b hover 同 rpcId 配对行高亮2 行)', pairedCount === 2, `paired=${pairedCount}`)
// §D-4 click a row to expand the JSON payload, click again to collapse
const firstRow = list.locator('button[class*="rowLine"]').first()
await firstRow.click()
const payloadShown = await list.locator('pre[class*="payload"]').count()
await firstRow.click()
const payloadHidden = await list.locator('pre[class*="payload"]').count()
report('§D-4 点行 JSON 展开/收起', payloadShown === 1 && payloadHidden === 0)
// §D-5 scrolling up pauses; resume restores follow
// First overflow the list (no scroll overflow → onScroll can never fire): click ping until ≥30 rows
while (await list.locator('button[class*="rowLine"]').count() < 30) {
await page.locator('button', { hasText: 'ping' }).click()
await page.waitForTimeout(30)
}
await list.evaluate((el) => { el.scrollTop = 0 })
await page.waitForSelector('div[class*="pausedBar"]', { timeout: 3000 })
const resumeBtn = page.locator('button', { hasText: '继续' })
report('§D-5 上滚触发暂停(按钮态+提示条)', await resumeBtn.count() === 1)
await resumeBtn.click()
await page.waitForTimeout(100)
const followRestored = await list.evaluate((el) => el.scrollHeight - el.scrollTop - el.clientHeight < 30)
report('§D-5b 继续恢复贴底跟随', followRestored)
// §D-6 clear: list empty, counters reset; periodic frames keep arriving (wait 6s for new rows)
await page.locator('button', { hasText: '清空' }).click()
const emptyAfterClear = await list.locator('button[class*="rowLine"]').count()
report('§D-6 清空后列表空', emptyAfterClear === 0)
await page.waitForFunction(
() => document.querySelectorAll('button[class*="rowLine"]').length > 0,
undefined, { timeout: 8000 },
)
report('§D-6b 清空后周期帧继续进入', true)
} catch (error) {
failures += 1
console.log(`FAIL 脚本异常 — ${error instanceof Error ? error.message : String(error)}`)
} finally {
await browser.close()
}
console.log(failures === 0 ? 'ALL PASS' : `${failures} FAILURE(S)`)
process.exit(failures === 0 ? 0 : 1)

View File

@@ -0,0 +1,132 @@
// Real-host spot check (condensed acceptance + connection stability): real sessions in the list,
// history renders on open, real prompt streams back. The stability assertions guard against
// fixture masking: fake streams never touch real SSE, so bridge-layer bugs (e.g. the req 'close'
// misdetection) only surface against a real host.
import { chromium } from 'playwright'
const BASE = process.env.VERIFY_BASE ?? 'http://127.0.0.1:3080'
let failures = 0
const report = (n, p, d = '') => { failures += p ? 0 : 1; console.log(`${p ? 'PASS' : 'FAIL'} ${n}${d ? ' — ' + d : ''}`) }
const browser = await chromium.launch()
try {
const page = await browser.newPage()
page.on('pageerror', (e) => console.log('[pageerror]', String(e).slice(0, 300)))
const apiRequests = []
let apiFailed = 0
page.on('request', (r) => { if (r.url().includes('/api/')) apiRequests.push(r.url()) })
page.on('requestfailed', (r) => { if (r.url().includes('/api/')) apiFailed++ })
await page.goto(`${BASE}/`, { waitUntil: 'load' })
// E2-0 connection stability: within a 12s window /api requests must be one-time setup cost
// (two streams + describe + list <= 10), zero aborts. The 300ms reconnect storm
// (the bridge bug fixed 2026-07-20) shows up here instantly.
await page.waitForTimeout(12000)
report('E2-0a 12s 内 /api 请求 ≤10无重连风暴', apiRequests.length <= 10, `count=${apiRequests.length}`)
report('E2-0b 无 requestfailedSSE 不被 client abort', apiFailed === 0, `failed=${apiFailed}`)
// E2-0c cold-session merge: list must include persisted sessions from previous host runs,
// not just in-memory attached ones (guards the R4 regression: first screen empty after
// restart). Requires at least one prior run's session on disk — every run of this script
// leaves some behind, so only a truly virgin .sessions root skips the assertion.
const listRes = await page.evaluate(async (base) => {
const res = await fetch(`${base}/api/session.list`, {
method: 'POST',
headers: { 'content-type': 'application/json' },
body: JSON.stringify({ type: 'client-request', rpcId: 'verify-cold-list', method: 'session.list', payload: {} }),
})
return res.json()
}, BASE)
const coldItems = listRes?.result?.ok ? listRes.result.value.items : []
const sorted = coldItems.every((it, i) => i === 0 || coldItems[i - 1].updatedAt >= it.updatedAt)
if (coldItems.length > 0) {
report('E2-0c 冷 session 进 list 且 updatedAt 倒序', sorted, `count=${coldItems.length}`)
const coldRows = await page.locator('aside button[class*="item"]').count()
report('E2-0d 首屏列表渲染冷 session非空', coldRows >= 1, `rows=${coldRows}`)
// Legacy no-cwd logs are not served (pre-release stance: no compatibility) —
// every listed session must carry its project cwd.
const noCwd = coldItems.filter((it) => typeof it.cwd !== 'string' || it.cwd.length === 0)
report('E2-0c2 无 cwd 存量不可见(全部条目携带 project cwd', noCwd.length === 0, `noCwd=${noCwd.length}`)
} else {
console.log('SKIP E2-0c/E2-0c2/E2-0d 冷 session 断言(.sessions 为空的全新 host')
}
// E2-0e error-channel fidelity: an unknown id must come back as session-not-found,
// never disguised as internal (and vice versa — guards the R3 regression).
const nf = await page.evaluate(async (base) => {
const res = await fetch(`${base}/api/session.history`, {
method: 'POST',
headers: { 'content-type': 'application/json' },
body: JSON.stringify({ type: 'client-request', rpcId: 'verify-not-found', method: 'session.history', payload: { sessionId: 'session-00000000-dead-beef-0000-000000000000' } }),
})
return res.json()
}, BASE)
report('E2-0e 未知 id 回 session-not-found不伪装 internal', nf?.result?.ok === false && nf.result.error.code === 'session-not-found', `code=${nf?.result?.error?.code}`)
// E2-1 create a real session into the list via '+' (covers the create path).
await page.locator('aside button[title="新建 session"]').click()
await page.waitForSelector('aside button[class*="item"]', { timeout: 8000 })
const n = await page.locator('aside button[class*="item"]').count()
report('E2-1 新建真 session 入列表', n >= 1, `count=${n}`)
// E2-1b default-project injection: a create without an explicit cwd must still get one
// (the host default — its process working directory), so the session lands in a project
// bucket instead of _no-cwd (guards the B-decision regression).
const afterCreate = await page.evaluate(async (base) => {
const res = await fetch(`${base}/api/session.list`, {
method: 'POST',
headers: { 'content-type': 'application/json' },
body: JSON.stringify({ type: 'client-request', rpcId: 'verify-default-cwd', method: 'session.list', payload: {} }),
})
return res.json()
}, BASE)
const newest = afterCreate?.result?.ok ? afterCreate.result.value.items[0] : undefined
report('E2-1b 新建 session 携带默认 cwdhost 进程目录注入)', typeof newest?.cwd === 'string' && newest.cwd.length > 0, `cwd=${newest?.cwd ?? '(absent)'}`)
// E2-2 open the first row: openState reaches open (input enabled)
await page.locator('aside button[class*="item"]').first().click()
await page.waitForSelector('main textarea:not([disabled])', { timeout: 8000 })
report('E2-2 打开真 sessionhistory 通、输入可用)', true)
// E2-3 real prompt: user bubble lands + partial pulse (real model streaming).
// Ask for a ~100-char reply: too-short replies finish inside waitForSelector's polling gap,
// making the pulse assertion race into a false failure.
await page.locator('main textarea').fill('用大约100字介绍事件溯源最后一句以「介绍完毕」结尾')
await page.locator('main button[class*="primary"]').click()
await page.waitForSelector('main div[class*="bubble"]', { timeout: 5000 })
report('E2-3a user 气泡入流', true)
const sawPulse = await page.waitForSelector('main span[class*="pulse"]', { timeout: 30000 }).then(() => true).catch(() => false)
report('E2-3b 真模型流式 partial 出现', sawPulse)
await page.waitForSelector('main span[class*="pulse"]', { state: 'detached', timeout: 60000 })
const text = (await page.locator('main').textContent()) ?? ''
report('E2-3c 回复定稿入流', text.includes('介绍完毕') || text.includes('事件溯源'), text.slice(-60))
// E2-4 stop mid-stream freezes the partial (aborted turns never finalize): the accumulated
// text survives as an interrupted terminal node (已停止 marker), the pulse stops, and later
// messages land after it. A reload must reconstruct the same node from the logged chunks.
const primary = page.locator('main button[class*="primary"]')
await page.locator('main textarea').fill('请从头背诵出师表全文,直接开始不要客套')
await primary.click()
await page.waitForSelector('main span[class*="pulse"]', { timeout: 30000 })
// Let visible content accumulate so the frozen node has a body to keep.
await page.waitForTimeout(2500)
await primary.click() // stop mid-stream (the no-finalize abort path)
await page.waitForSelector('main div[class*="head"] span[data-running]', { state: 'detached', timeout: 15000 })
const pulseGone = await page.waitForSelector('main span[class*="pulse"]', { state: 'detached', timeout: 2000 }).then(() => true).catch(() => false)
const frozenMark = await page.locator('main span[class*="stopped"]', { hasText: '已停止' }).count()
report('E2-4a 停止后 partial 定格(脉冲停+已停止标记+文本保留)', pulseGone && frozenMark >= 1, `pulseGone=${pulseGone} marks=${frozenMark}`)
await page.locator('main textarea').fill('请只回复四个字:顺序正常')
await primary.click()
await page.waitForSelector('main div[class*="bubble"]:has-text("顺序正常")', { timeout: 10000 })
const rows = await page.evaluate(() => {
const scroll = document.querySelector('main div[class*="scroll"]')
return [...(scroll?.children ?? [])].map((el) => (el.textContent ?? '').trim()).filter(Boolean)
})
const iStopped = rows.findIndex((t) => t.includes('出师表'))
const iNew = rows.findIndex((t) => t.includes('顺序正常'))
report('E2-4b 停止后再发消息顺序正确(新消息在末尾)', iNew > iStopped && iStopped >= 0, `stopped@${iStopped} new@${iNew}`)
// E2-4c reload: history replay re-freezes the interrupted node (live view and replay agree).
await page.waitForSelector('main span[class*="pulse"]', { state: 'detached', timeout: 60000 })
await page.reload({ waitUntil: 'load' })
await page.locator('aside button[class*="item"]').first().click()
await page.waitForSelector('main textarea:not([disabled])', { timeout: 8000 })
const marksAfterReload = await page.locator('main span[class*="stopped"]', { hasText: '已停止' }).count()
report('E2-4c 刷新后中断消息仍在history 重建一致)', marksAfterReload >= 1, `marks=${marksAfterReload}`)
} catch (e) {
failures += 1
console.log(`FAIL 脚本异常 — ${String(e).slice(0, 300)}`)
} finally {
await browser.close()
}
console.log(failures === 0 ? 'ALL PASS' : `${failures} FAILURE(S)`)
process.exit(failures === 0 ? 0 : 1)

View File

@@ -0,0 +1,361 @@
// Session UI browser acceptance (fixture mode); step tags match the report labels.
// Prereqs: dsh web running on 3080, apps/web/dist freshly built, playwright chromium installed.
// Run: node missions/scripts/verify-session.mjs (not part of any gate system)
import { chromium } from 'playwright'
const BASE = process.env.DSH_WEB_URL ?? 'http://127.0.0.1:3080'
let failures = 0
function report(name, pass, detail = '') {
failures += pass ? 0 : 1
console.log(`${pass ? 'PASS' : 'FAIL'} ${name}${detail ? `${detail}` : ''}`)
}
const browser = await chromium.launch()
try {
const page = await browser.newPage()
await page.goto(`${BASE}/?fixture`, { waitUntil: 'load' })
// §E1-1 three list rows + fx-alpha running dot + fx-beta lineage indent + empty right pane
await page.waitForSelector('aside button[class*="item"]', { timeout: 5000 })
const items = page.locator('aside button[class*="item"]')
report('§E1-1a 列表 3 条', await items.count() === 3, `count=${await items.count()}`)
const alphaDot = page.locator('aside button[title="fx-alpha"] span[class*="running"]')
report('§E1-1b fx-alpha running 绿点', await alphaDot.count() === 1)
const betaPad = await page.locator('aside button[title="fx-beta"]').evaluate((el) => el.style.paddingLeft)
report('§E1-1c fx-beta 谱系缩进depth=1 → 24px', betaPad === '24px', `paddingLeft=${betaPad}`)
report('§E1-1d 右侧空态', (await page.locator('main').textContent())?.includes('选择或新建') ?? false)
// §E1-2 open fx-alpha: all history node kinds render, scroll lands at bottom
await page.locator('aside button[title="fx-alpha"]').click()
await page.waitForSelector('main div[class*="bubble"]', { timeout: 5000 })
const mainText = await page.locator('main').textContent()
report('§E1-2a user 气泡渲出', (mainText ?? '').includes('问题 59'))
report('§E1-2b assistant 正文渲出', (mainText ?? '').includes('回答 59'))
// Bottom check BEFORE expanding reasoning (a local expand grows height without triggering follow — view state, not a snapshot change; by design).
const scroll = page.locator('main div[class*="scroll"]')
const atBottom = await scroll.evaluate((el) => el.scrollHeight - el.scrollTop - el.clientHeight < 30)
report('§E1-2i 打开后滚动在底部', atBottom)
const reasoningToggle = page.locator('main button[class*="reasoningToggle"]').last()
report('§E1-2c reasoning 折叠钮存在', await reasoningToggle.count() > 0)
await reasoningToggle.click()
report('§E1-2d reasoning 展开有内容', ((await page.locator('main').textContent()) ?? '').includes('思考过程'))
report('§E1-2e 工具卡渲出', await page.locator('main div[class*="card"] span[class*="name"]', { hasText: 'echo' }).count() > 0)
report('§E1-2f steering 徽标渲出', await page.locator('main span[class*="badge"]', { hasText: '插话' }).count() > 0)
report('§E1-2g context 折叠卡渲出', await page.locator('main button', { hasText: '上下文注入' }).count() > 0)
report('§E1-2h 常驻审批占位卡', await page.locator('main div[class*="card"]', { hasText: '等待审批' }).count() === 1)
// §E1-3 load-older: prepend one page, viewport stays anchored
const olderBtn = page.locator('main button', { hasText: '加载更早' })
report('§E1-3a hasMore 显示加载更早钮', await olderBtn.count() === 1)
const beforeAnchor = await scroll.evaluate((el) => ({ h: el.scrollHeight, t: el.scrollTop }))
await scroll.evaluate((el) => { el.scrollTop = 0 }) // scroll up before paging (realistic gesture)
const anchorTop = await scroll.evaluate((el) => el.scrollTop)
await olderBtn.click()
await page.waitForFunction((prev) => {
const el = document.querySelector('main div[class*="scroll"]')
return el !== null && el.scrollHeight > prev
}, beforeAnchor.h, { timeout: 5000 })
const afterAnchor = await scroll.evaluate((el) => ({ h: el.scrollHeight, t: el.scrollTop }))
const drift = Math.abs(afterAnchor.t - (anchorTop + (afterAnchor.h - beforeAnchor.h)))
report('§E1-3b 翻页锚定scrollTop 补偿高度差)', drift < 4, `drift=${drift}px`)
report('§E1-3c 更早消息已前插', ((await page.locator('main').textContent()) ?? '').includes('问题 20'))
// §E1-5 send (queue): user bubble lands + typewriter partial + finalize; draft clears.
// Button rulings 2026-07-20: one primary button (send idle / stop running); running locks the input.
const input = page.locator('main textarea')
const primaryBtn = page.locator('main button[class*="primary"]')
// fx-alpha opens running=true (fixture list material) — the merged primary reads 停止 there; reset to idle first.
if (await primaryBtn.getAttribute('aria-label') === '停止') {
await primaryBtn.click()
await page.waitForSelector('main div[class*="head"] span[data-running]', { state: 'detached', timeout: 5000 })
}
await input.fill('验收消息一')
await primaryBtn.click()
await page.waitForSelector('main div[class*="bubble"]:has-text("验收消息一")', { timeout: 3000 })
report('§E1-5a user 气泡入流', true)
report('§E1-5b 草稿清空', await input.inputValue() === '')
// Typewriter: the partial pulse is visible
await page.waitForSelector('main span[class*="pulse"]', { timeout: 3000 })
report('§E1-5c 流式 partial 脉冲出现', true)
// Running dot lights up (fixture prompt flips status)
await page.waitForSelector('main div[class*="head"] span[data-running]', { timeout: 3000 })
report('§E1-5d running 状态点亮', true)
// §E1-6 running locks the input (ruling 2026-07-20 #3, supersedes the hover menu):
// textarea disabled (draft visible but frozen), no queue/steer menu, stop is the only action.
report('§E1-6a running 时输入框置灰', await input.isDisabled())
report('§E1-6b running 时无排队/插话菜单', await page.locator('main button[class*="menuItem"]').count() === 0)
report('§E1-6c running 时主按钮为停止且可用', await primaryBtn.isEnabled() && (await primaryBtn.getAttribute('aria-label')) === '停止')
// Wait for finalize: pulse gone + echo body present (partial -> finalized node swap)
await page.waitForSelector('main span[class*="pulse"]', { state: 'detached', timeout: 15000 })
report('§E1-5e 定稿切换(脉冲消失)', true)
report('§E1-5f 回声正文定稿', ((await page.locator('main').textContent()) ?? '').includes('回声:验收消息一'))
// §E1-7 stop: send another, the primary button flips to stop (same slot) mid-replay
await input.fill('验收消息二')
await primaryBtn.click()
await page.waitForSelector('main button[aria-label="停止"]', { timeout: 3000 })
report('§E1-7d 运行中主按钮原地变停止', true)
await primaryBtn.click() // now the stop action
await page.waitForSelector('main div[class*="head"] span[data-running]', { state: 'detached', timeout: 5000 })
report('§E1-7a 停止后 running 熄灭', true)
report('§E1-7b 中断标记入流', ((await page.locator('main').textContent()) ?? '').includes('(已中断)'))
report('§E1-7e 停止后主按钮回到发送', await primaryBtn.getAttribute('aria-label') === '发送')
// Turn end unlocks the box and returns focus (before any fill taints activeElement).
await page.waitForTimeout(200)
report('§E1-7f 停止解禁后焦点回输入框', await page.evaluate(() => document.activeElement?.tagName === 'TEXTAREA'))
await input.fill('x')
await primaryBtn.hover()
await page.waitForTimeout(300)
report('§E1-7c 无排队/插话菜单(空闲 hover 亦无)', await page.locator('main button[class*="menuItem"]').count() === 0)
await input.fill('')
// §E1-8 switch to fx-beta and back: empty conversation / instant re-render (resident instances)
await page.locator('aside button[title="fx-beta"]').click()
await page.waitForFunction(() => {
const main = document.querySelector('main')
return main !== null && (main.textContent ?? '').includes('fx-beta')
}, undefined, { timeout: 3000 })
const betaBubbles = await page.locator('main div[class*="bubble"]').count()
report('§E1-8a fx-beta 空对话', betaBubbles === 0, `bubbles=${betaBubbles}`)
const t0 = Date.now()
await page.locator('aside button[title="fx-alpha"]').click()
await page.waitForSelector('main div[class*="bubble"]:has-text("验收消息一")', { timeout: 2000 })
report('§E1-8b 切回 fx-alpha 即时呈现(常驻实例)', Date.now() - t0 < 1500, `${Date.now() - t0}ms`)
// §E1-9 create selects and opens immediately
const before = await items.count()
await page.locator('aside button[title="新建 session"]').click()
await page.waitForFunction((n) => document.querySelectorAll('aside button[class*="item"]').length > n, before, { timeout: 3000 })
const newSelected = await page.locator('aside button[class*="selected"]').getAttribute('title')
report('§E1-9 新建即入列表并选中', newSelected !== null && newSelected.startsWith('fx-'), `selected=${newSelected}`)
// §E1-11 InputBar regression pins (IME composition / autorepeat / caret / autosize / draft semantics)
await page.locator('aside button[title="fx-alpha"]').click()
const inputBox = page.locator('main textarea')
await inputBox.waitFor({ timeout: 3000 })
// B1: composition Enter must not send (IME candidate pick)
const bubblesB1 = await page.locator('main div[class*="bubble"]').count()
await inputBox.fill('IME 探测')
await inputBox.evaluate((el) => {
el.dispatchEvent(new KeyboardEvent('keydown', { key: 'Enter', keyCode: 229, isComposing: true, bubbles: true, cancelable: true }))
})
await page.waitForTimeout(200)
report('§E1-11a IME 组合期 Enter 不发送', await page.locator('main div[class*="bubble"]').count() === bubblesB1 && await inputBox.inputValue() === 'IME 探测')
// B6: key-repeat Enter must not send
await inputBox.evaluate((el) => {
el.dispatchEvent(new KeyboardEvent('keydown', { key: 'Enter', bubbles: true, cancelable: true, repeat: true }))
})
await page.waitForTimeout(200)
report('§E1-11b Enter 长按 autorepeat 不发送', await page.locator('main div[class*="bubble"]').count() === bubblesB1)
await inputBox.fill('')
// B2: mid-text edit keeps the caret (synchronous controlled-value notify)
await inputBox.fill('abcdef')
await inputBox.evaluate((el) => el.setSelectionRange(3, 3))
await inputBox.press('x')
await page.waitForTimeout(100)
const caret = await inputBox.evaluate((el) => ({ v: el.value, s: el.selectionStart }))
report('§E1-11c 中段编辑光标不跳', caret.v === 'abcxdef' && caret.s === 4, `value=${caret.v} caret=${caret.s}`)
await inputBox.fill('')
// B3: soft-wrap long text grows the box (mirror-div auto-grow), capped at the 14-line baseline (336px)
const hEmpty = (await inputBox.boundingBox())?.height ?? 0
await inputBox.fill('这是一段没有换行符但是非常长的文本'.repeat(60))
await page.waitForTimeout(100)
const hLong = (await inputBox.boundingBox())?.height ?? 0
report('§E1-11d 软换行自增高且封顶', hLong > hEmpty + 20 && hLong <= 344, `h ${hEmpty} -> ${hLong}`)
await inputBox.fill('')
// B4 (reworked under ruling 3): sending locks the box for the turn; focus returns on unlock (pinned at §E1-7f)
const primary = page.locator('main button[class*="primary"]')
await inputBox.fill('焦点验收')
await primary.click()
await page.waitForTimeout(200)
report('§E1-11e 发送后运行期输入锁定', await inputBox.isDisabled())
report('§E1-11f 发送即清稿(乐观清)', await inputBox.inputValue() === '')
// B5: single primary slot — the button must not move when running flips (send<->stop in place)
await page.waitForSelector('main button[aria-label="停止"]', { timeout: 3000 })
const yRunning = (await primary.boundingBox())?.y ?? -1
await primary.click() // stop
await page.waitForSelector('main div[class*="head"] span[data-running]', { state: 'detached', timeout: 5000 })
const yIdle = (await primary.boundingBox())?.y ?? -2
report('§E1-11g 发送/停止原地切换不跳动', Math.abs(yRunning - yIdle) < 2, `primary.y ${yRunning} vs ${yIdle}`)
// Sending force-scrolls to the bottom even when scrolled away (own words must be visible;
// passive follow still respects scrolled-away readers during streaming).
const scrollBox = page.locator('main div[class*="scroll"]')
await scrollBox.evaluate((el) => { el.scrollTop = 0 })
await inputBox.fill('置底验收消息')
await primary.click()
await page.waitForSelector('main div[class*="bubble"]:has-text("置底验收消息")', { timeout: 3000 })
const nearBottom = await scrollBox.evaluate((el) => el.scrollHeight - el.scrollTop - el.clientHeight < 30)
report('§E1-11h 上滚状态下发送强制置底', nearBottom)
await page.waitForSelector('main button[aria-label="停止"]', { timeout: 3000 })
await primary.click() // stop the replay to leave the fixture idle
await page.waitForSelector('main div[class*="head"] span[data-running]', { state: 'detached', timeout: 5000 })
// §E1-10 RPC panel cross-check: this run's traffic is visible in the ledger (history/prompt/cancel round trips).
// The ledger is a left-menu bar page now: activate it from the icon rail, assert panel-area content.
await page.locator('nav button[title="RPC 日志"]').click()
await page.waitForSelector('section[class*="panel"]', { timeout: 3000 })
const panelText = (await page.locator('section[class*="panel"]').textContent()) ?? ''
const sawHistory = panelText.includes('session.history') || panelText.includes('session/event')
report('§E1-10 调试面板见 session 流量', sawHistory)
await page.locator('nav button[title="会话列表"]').click() // restore the sessions panel for later steps
// §E1-12 时序批audit S1/S3/S4fixture __fxTiming 后门制造慢 history / 丢帧 / 重连窗口。
// 新开 page = 全新 fixture 实例,不受上面步骤污染。
const page2 = await browser.newPage()
await page2.goto(`${BASE}/?fixture`, { waitUntil: 'load' })
await page2.waitForSelector('aside button[title="fx-alpha"]', { timeout: 5000 })
// S1: open 窗口期来的 live 帧必须缝合进窗口(慢 history 下打开正在流式的会话)
await page2.evaluate(() => globalThis.__fxTiming.setHistoryDelay(700))
await page2.locator('aside button[title="fx-alpha"]').click()
await page2.waitForTimeout(120) // open 在途history 还有 ~580ms 才回)
await page2.evaluate(() => {
globalThis.__fxTiming.appendUser('fx-alpha', '开窗期实时消息A')
globalThis.__fxTiming.appendUser('fx-alpha', '开窗期实时消息B')
})
// 就绪信号用气泡而非 textareafx-alpha 初始 running=true输入框在 running 期被禁用
await page2.waitForSelector('main div[class*="bubble"]', { timeout: 8000 })
await page2.waitForSelector('main div[class*="bubble"]:has-text("开窗期实时消息B")', { timeout: 5000 }).catch(() => {})
await page2.evaluate(() => globalThis.__fxTiming.setHistoryDelay(0))
const s1Text = (await page2.locator('main').textContent()) ?? ''
report('§E1-12a open 期间来帧缝合不丢S1', s1Text.includes('开窗期实时消息A') && s1Text.includes('开窗期实时消息B'))
report('§E1-12b 缝合后无 fold 降级S1', !s1Text.includes('历史视图降级'))
// S3: 途中丢一帧造出 seq 洞 → resync-lite 重拉尾页找回丢帧,且不触发 fold 降级
await page2.evaluate(() => {
globalThis.__fxTiming.appendSilent('fx-alpha', '途中丢失的消息C') // 只进 log 不发 mux 帧
globalThis.__fxTiming.appendUser('fx-alpha', '洞后到达的消息D') // client 看见 seq 跳 2
})
const gapRepaired = await page2.waitForSelector('main div[class*="bubble"]:has-text("途中丢失的消息C")', { timeout: 5000 }).then(() => true).catch(() => false)
report('§E1-12c seq 洞触发补拉,丢帧经 history 找回S3', gapRepaired)
const s3Text = (await page2.locator('main').textContent()) ?? ''
report('§E1-12d 洞后帧不丢S3', s3Text.includes('洞后到达的消息D'))
report('§E1-12e 洞不再触发 fold 降级S3', !s3Text.includes('历史视图降级'))
// S4: open 在途时断线,在途 history 注定失败 → 重连 resync 的 generation 必须作废旧结果,
// 不得在新窗口成功打开后被迟到的旧失败定格成 error。用 fx-beta无定时素材running 恒 false
await page2.evaluate(() => {
globalThis.__fxTiming.setHistoryDelay(1200)
globalThis.__fxTiming.failNextHistory()
})
await page2.locator('aside button[title="fx-beta"]').click()
await page2.waitForTimeout(150) // 注定失败的 open 在途
await page2.evaluate(() => {
globalThis.__fxTiming.setHistoryDelay(0)
globalThis.__fxTiming.breakStreams() // 双流断 → 重连(退避 ~250-500ms + 宽限 150ms→ resync
})
await page2.waitForSelector('main textarea:not([disabled])', { timeout: 8000 })
await page2.waitForTimeout(1400) // 等旧 doomed 请求(~1350ms 处)失败落地后再断言
const s4ErrStrips = await page2.locator('main div[class*="openError"]').count()
const s4InputOk = await page2.locator('main textarea:not([disabled])').count()
report('§E1-12f 断线窗口在途 open 不定格失败S4 generation 作废)', s4ErrStrips === 0 && s4InputOk === 1, `errStrips=${s4ErrStrips} input=${s4InputOk}`)
await page2.close()
// §E1-13 引用稳定audit S5+C3流式 chunk 期间 memo 必须真实命中——
// 稳定的 ToolCallCard/SessionListItem 渲染次数不随 chunk 帧线性增长。
const page3 = await browser.newPage()
await page3.addInitScript(() => { globalThis.__renderCounts = {} })
await page3.goto(`${BASE}/?fixture`, { waitUntil: 'load' })
await page3.waitForSelector('aside button[title="fx-alpha"]', { timeout: 5000 })
await page3.locator('aside button[title="fx-alpha"]').click()
await page3.waitForSelector('main div[class*="bubble"]', { timeout: 8000 })
// 复位到空闲fx-alpha 开局 running
const primary3 = page3.locator('main button[class*="primary"]')
if (await primary3.getAttribute('aria-label') === '停止') {
await primary3.click()
await page3.waitForSelector('main div[class*="head"] span[data-running]', { state: 'detached', timeout: 5000 })
}
await page3.evaluate(() => { globalThis.__renderCounts = {} })
// 发送触发 fixture 流式回放(约 20+ 个 chunk 帧)
await page3.locator('main textarea').fill('memo 稳定性验收')
await primary3.click()
await page3.waitForSelector('main span[class*="pulse"]', { timeout: 5000 })
await page3.waitForSelector('main span[class*="pulse"]', { state: 'detached', timeout: 20000 })
const counts = await page3.evaluate(() => globalThis.__renderCounts)
// 历史窗口 50 条消息里有 ~10 张工具卡全部已定稿chunk 期间它们的 props 引用应稳定,
// memo 全程命中 → 整轮流式回放中每张卡渲染次数为 0发送时快照 nodes 未变)。
// 列表条目running 翻转 2 次true/false+ updatedAt 变 1 次是合法渲染,帧驱动重渲则会到几十次。
const toolRenders = counts.ToolCallCard ?? 0
const listRenders = counts.SessionListItem ?? 0
report('§E1-13a 流式期间已定稿工具卡 memo 命中S5', toolRenders <= 12, `ToolCallCard renders=${toolRenders}10 卡;>12 即 memo 失效)`)
report('§E1-13b 流式期间列表条目 memo 命中S5+C3', listRenders <= 12, `SessionListItem renders=${listRenders}3 行 × 合法状态翻转;>12 即 memo 失效)`)
// §E1-15 tool 卡三级回退toolcard-wirefixture 60-62 turn 携带三型 view 样本;
// echo无 presenter钉住无 view 兜底 JSON 卡路径。
{
const scroll3 = page3.locator('main div[class*="scroll"]')
// fx-bash terminal 卡:命令占卡头 name 槽 + cwd + exit 胶囊 + 输出
const termCmd = await page3.locator('main span[class*="name"]', { hasText: 'ls -la' }).count()
const termCwd = await page3.locator('main span[class*="cwd"]', { hasText: '/tmp/fixture' }).count()
const termPill = await page3.locator('main span[class*="pill"]', { hasText: 'exit 0' }).count()
report('§E1-15a terminal 卡渲出(命令+cwd+exit 胶囊)', termCmd >= 1 && termCwd >= 1 && termPill >= 1, `cmd=${termCmd} cwd=${termCwd} pill=${termPill}`)
const termOut = (await scroll3.textContent() ?? '').includes('drwxr-xr-x fixture')
report('§E1-15b terminal 卡输出体渲出', termOut)
// fx-write diff 卡path 头 + 新文本块
const diffPath = await page3.locator('main div[class*="diffPath"]', { hasText: 'notes/demo.txt' }).count()
const diffNew = await page3.locator('main pre[class*="diffNew"]', { hasText: 'hello fixture' }).count()
report('§E1-15c diff 卡渲出path 头+新文本)', diffPath >= 1 && diffNew >= 1, `path=${diffPath} new=${diffNew}`)
// fx-note generic 卡view 标题上头 + kind 图标
const genTitle = await page3.locator('main span[class*="name"]', { hasText: '记录笔记' }).count()
report('§E1-15d generic 卡渲出view 标题)', genTitle >= 1, `title=${genTitle}`)
// echo 无 presenter老 JSON 折叠卡兜底(参数折叠钮仍在)
const echoCard = await page3.locator('main div[class*="card"]:has(span[class*="name"]:text-is("echo")) button', { hasText: '参数' }).count()
report('§E1-15e 无 view 工具兜底 JSON 卡echo', echoCard >= 1, `echoParamToggles=${echoCard}`)
}
// §E1-16 壳骨架app-shell knife 3tabs 条在、占位页渲、点 tool 卡展开右栏 detail、再点收起。
{
// tabs 条conversation + gantt 两 tab 注册后条自然出现(单 tab 时不渲染的分支反证)。
const tabConv = await page3.locator('main button', { hasText: '会话' }).count()
const tabGantt = await page3.locator('main button', { hasText: '甘特' }).count()
report('§E1-16a tabs 条渲出(会话+甘特)', tabConv >= 1 && tabGantt >= 1, `conv=${tabConv} gantt=${tabGantt}`)
// 占位页:切甘特 tab 渲说明性占位,切回会话流还在。
await page3.locator('main button', { hasText: '甘特' }).click()
const placeholderText = (await page3.locator('main').textContent()) ?? ''
report('§E1-16b 甘特占位页渲出', placeholderText.includes('视图建设中'))
await page3.locator('main button', { hasText: '会话' }).first().click()
await page3.waitForSelector('main div[class*="bubble"]', { timeout: 3000 })
// 点 tool 卡头 → 右栏 detail 展开callId+argsRaw JSON同卡再点 → 收起。
const echoHead = page3.locator('main div[class*="card"]:has(span[class*="name"]:text-is("echo")) div[class*="head"]').first()
await echoHead.click()
const detailShown = await page3.waitForSelector('span[class*="detailTitle"]', { timeout: 3000 }).then(() => true).catch(() => false)
const detailText = detailShown ? (await page3.locator('div[class*="detailBody"]').textContent()) ?? '' : ''
report('§E1-16c 点卡展开右栏 detailcallId+argsRaw', detailShown && detailText.includes('callId') && detailText.includes('argsRaw'))
await echoHead.click()
const detailGone = await page3.waitForSelector('span[class*="detailTitle"]', { state: 'detached', timeout: 3000 }).then(() => true).catch(() => false)
report('§E1-16d 同卡再点收起', detailGone)
// 关闭钮路径:再开一次,点 × 收起(空 detail 兜底由 jsdom 层守)。
await echoHead.click()
await page3.waitForSelector('span[class*="detailTitle"]', { timeout: 3000 })
await page3.locator('button[title="关闭详情"]').click()
const closedByBtn = await page3.waitForSelector('span[class*="detailTitle"]', { state: 'detached', timeout: 3000 }).then(() => true).catch(() => false)
report('§E1-16e 关闭钮收起', closedByBtn)
}
// §E1-14 连接状态可见audit C1断流 → 顶部细条出现;重连成功 → 细条消失。
report('§E1-14a 连接正常时无断线细条', await page3.locator('div[class*="banner"]').count() === 0)
await page3.evaluate(() => globalThis.__fxTiming.breakStreams())
const bannerShown = await page3.waitForSelector('div[class*="banner"]', { timeout: 5000 }).then(() => true).catch(() => false)
report('§E1-14b 断流后重连细条出现', bannerShown)
const bannerGone = await page3.waitForSelector('div[class*="banner"]', { state: 'detached', timeout: 8000 }).then(() => true).catch(() => false)
report('§E1-14c 重连成功后细条消失', bannerGone)
await page3.close()
} catch (error) {
failures += 1
console.log(`FAIL 脚本异常 — ${error instanceof Error ? error.message : String(error)}`)
} finally {
await browser.close()
}
console.log(failures === 0 ? 'ALL PASS' : `${failures} FAILURE(S)`)
process.exit(failures === 0 ? 0 : 1)

View File

@@ -0,0 +1,45 @@
// Webserver bridge backpressure probe (audit R2, webserver half): a paused client socket
// must stall the SSE pump (res.write false → await drain) instead of buffering unboundedly.
// Self-contained: starts startWebServer on PORT with a stub apiHandler; no host needed.
// Run: node_modules/.bin/tsx missions/scripts/verify-webserver-backpressure.mjs
import { connect } from 'node:net'
import { once } from 'node:events'
import { startWebServer } from '../../packages/host/webserver/src/index.ts'
const PORT = Number(process.env.PROBE_PORT ?? 3097)
const CHUNK = 64 * 1024
const TOTAL = 200 // 200 × 64KB = 12.5MB — far beyond any socket buffer
let failures = 0
const report = (n, p, d = '') => { failures += p ? 0 : 1; console.log(`${p ? 'PASS' : 'FAIL'} ${n}${d ? ' — ' + d : ''}`) }
let pulled = 0
const apiHandler = {
fetch: async () => new Response(new ReadableStream({
pull(controller) {
if (pulled >= TOTAL) return controller.close()
pulled++
controller.enqueue(new Uint8Array(CHUNK))
},
}), { headers: { 'content-type': 'text/event-stream' } }),
}
const server = await startWebServer({ port: PORT, distIndex: '/nonexistent/index.html', apiHandler }, (e) => console.error(String(e)))
const socket = connect(PORT, '127.0.0.1')
await once(socket, 'connect')
socket.write(`GET /api/events.host HTTP/1.1\r\nHost: x\r\nConnection: keep-alive\r\n\r\n`)
socket.pause() // stop reading: kernel+node buffers fill, then res.write must return false
await new Promise(r => setTimeout(r, 1500))
const stalled = pulled
// Without drain-await the pump races through all chunks regardless of the paused reader.
report('R2 暂停读的慢客户端使泵停在低水位(非全量吞入内存)', stalled < TOTAL / 2, `pulled=${stalled}/${TOTAL}`)
socket.resume() // drain: the pump must resume and finish
const t0 = Date.now()
while (pulled < TOTAL && Date.now() - t0 < 10_000) await new Promise(r => setTimeout(r, 100))
report('R2b 恢复读后泵继续推进到完成', pulled === TOTAL, `pulled=${pulled}/${TOTAL}`)
socket.destroy()
await server.close()
console.log(failures === 0 ? 'ALL PASS' : `${failures} FAILURE(S)`)
process.exit(failures === 0 ? 0 : 1)

View File

@@ -0,0 +1,37 @@
// Webserver hardening probe (audit R1): malformed requests must yield 4xx/5xx, never kill
// the process. Prereq: dsh web on 3080. Run: node missions/scripts/verify-webserver-hardening.mjs
const BASE = process.env.DSH_WEB_URL ?? 'http://127.0.0.1:3080'
let failures = 0
function report(name, pass, detail = '') {
failures += pass ? 0 : 1
console.log(`${pass ? 'PASS' : 'FAIL'} ${name}${detail ? `${detail}` : ''}`)
}
async function status(path, init) {
try {
return (await fetch(`${BASE}${path}`, init)).status
} catch {
return 0 // connection refused/reset — the server died or dropped us
}
}
// R1 trigger set: bad %-encodings (decodeURIComponent URIError), long path, bad method, non-JSON API body.
const cases = [
['/%', [400]],
['/%c0', [400]],
['/%zz%', [400]],
['/' + 'a'.repeat(9000), [200]], // SPA fallback, must not throw
['/foo', [405], { method: 'DELETE' }],
['/api/session.list', [400], { method: 'POST', body: 'not json' }],
]
for (const [path, expect, init] of cases) {
const got = await status(path, init)
const label = path.length > 24 ? `${path.slice(0, 24)}` : path
report(`${init?.method ?? 'GET'} ${label} -> ${expect.join('/')}`, expect.includes(got), `got=${got}`)
}
report('server alive after the barrage (GET / -> 200)', (await status('/')) === 200)
console.log(failures === 0 ? 'ALL PASS' : `${failures} FAILURE(S)`)
process.exit(failures === 0 ? 0 : 1)