mlx-bun 0.2.0 → 0.3.0

package/docs/design/web-chat-redesign.md

@@ -1,1112 +1,695 @@
-# Web chat redesign: the definitive plan
-
-Status: proposed · Owner: web/chat surface · Scope: `src/web/app.html`, `src/pi-web.ts`, `src/server.ts` (chat routes), `src/memory/*`, `src/lora.ts` · Horizon: one quarter, phased
+---
+status: active
+axis: USING
+canonical-for: web-chat
+plan-anchor: none
+last-verified: 2026-08-23
+---
 
-This document merges three independent design visions, ten market-research
-briefs, and a line-level audit of the current UI and backend into one
-execution plan. It supersedes `docs/archive/planning/chat-ui-vision.md` and
-`docs/archive/planning/chat-ui-vision-memory-first.md` as the working plan (keep
-those two as historical inputs; this is the doc that gets executed against).
-It also absorbs and re-prioritizes every item in `docs/archive/planning/web-ui-pass-plan.md`
-— those bug-fixes are folded into Phase 0/1 below by number.
+# Web chat — the built-in chat surface (canonical design)
+
+Scope: `src/web/` (app.html + `src/web/src/*.ts` built into `app.js`),
+`src/pi-web.ts` (the WebSocket bridge), the thin REST wrappers the UI calls
+(`src/memory/rest.ts`, `src/hub-rest.ts`, `src/serve/session-search.ts`,
+`src/serve/model-admin-routes.ts`), `src/tool-approvals.ts`.
+
+This is the one canonical doc for the web chat. It absorbs the former
+beat matrix (`docs/design/web-chat-redesign.md`, kept as Appendix A), the web half
+of `docs/design/web-chat-redesign.md` (§3.6 here), and the desktop-shell idea (one
+History line). Status/changelog prose belongs in PLAN.md; this file holds
+the design, the as-built contract, and the open items.
+
+**Standing decision (2026-06-14): the chat UI is a Pi embed, not a
+hand-rolled chat.** The browser talks to an in-process
+`@earendil-works/pi-coding-agent` `AgentSession` over `/ws/chat`
+(`src/pi-web.ts`). Pi owns the agent loop, tool execution, session files
+(new/list/open/fork/delete, `SessionManager.forkFrom`), the tree
+navigation behind edit/regenerate (`AgentSession.navigateTree`),
+compaction, thinking levels, and the steering/follow-up queue. Nothing in
+this design rebuilds those; the web work is (a) frames that expose them,
+(b) UI that makes them legible, (c) the few capabilities Pi has no
+concept of (memory vault, LoRA adapters, serving lanes, hardware fit,
+the app-aware assistant).
 
 ---
 
-## 1. Executive summary
-
-**The thesis:** mlx-bun's web chat should not try to be a better ChatGPT. It
-should be the only local chat surface where the AI visibly, provably gets
-better at knowing *you*, where switching to a domain-specific brain is a
-first-class routing decision instead of a filename in a dropdown, and where
-"local and private" is a proof you can see on screen, not a claim in a
-README. Every serious competitor surveyed — Ollama, LM Studio, Open WebUI,
-Odysseus, Jan, Msty, macMLX — either has no personalization story at all
-(the MLX-native peers, Ollama, LM Studio) or bolts memory on as a flat
-prompt-stuffed fact list (Open WebUI, AnythingLLM, ChatGPT, Claude). None of
-them combine automatic personalization **with** weight-level adaptation
-(LoRA hot-swap) **with** native MLX serving speed **with** a legible,
-auditable tool-trust model. That combination is mlx-bun's only defensible
-wedge, and today the UI hides it: the backend (`src/pi-web.ts`, `src/memory/tools.ts`,
-`src/lora.ts`, `src/server.ts`) already implements most of it; the 2737-line
-`src/web/app.html` shows almost none of it.
-
-**The strategic bet (revised 2026-07-06):** best-in-class is a **superset,
-not a wedge**. The goal is the single best UI for macOS local AI, period —
-so the qualification round is conceding *nothing*: for every capability a
-user judges a local chat app by, we match or beat the best current
-implementation, or we make a written skip decision with a substrate reason.
-The acceptance test: nobody ever says "I miss how Ollama / LM Studio /
-Open WebUI / optiq had X." The axis-by-axis requirement table lives in
-[web-chat-beat-matrix.md](web-chat-beat-matrix.md) (12 axes, 116 capability
-rows, each MATCH / BEAT / SKIP-with-reason) — it is the coverage checklist
-every phase below is planned against. *(An earlier draft of this section
-framed the bet as "don't chase Open WebUI's breadth or LM Studio's polish";
-that conflated where the moat is with what to build, and is reversed.)*
-
-**On top of that conceded-nothing surface**, we stack the three things no
-competitor's architecture allows them to show: (1) memory with a vault you
-can open, search, and watch self-heal after a correction — today that is
-git-tracked articles + nightly synthesis (real, shipped); the "graduates
-into a trained memory LoRA" rung is roadmap, not current capability (the
-adapter runtime exists, the memory fine-tunes don't yet — the UI must never
-claim it before it ships); (2) LoRA routing as a legible per-surface
-decision, not a Python-API-only mechanism (optiq built the engine, never
-the UI; LM Studio's own bug tracker confirms adapter hot-swap is
-unsupported industry-wide); (3) serving speed and tool trust as ambient,
-always-on proof — tok/s, TTFT, which lane is active (serial / spec-decode /
-batched), and exactly what a tool call touched on disk — because we control
-the whole stack from Metal kernel to composer pixel and nobody else
-surveyed does. The differentiators are the reason to *switch*; they are
-never an excuse for a missing table stake.
-
-**The pragmatic constraint:** almost none of this requires new backend
-capability. Session fork, tool approval, adapter-scoped KV, the memory tool
-suite, structured output, the sampler extension set — all of it exists
-server-side today with zero or thin UI. The fastest, highest-leverage path
-is wiring existing capability to new UI, not a rewrite. Where a rewrite *is*
-warranted (the streaming-render architecture, message-tree branching,
-virtualized long threads), the plan says so explicitly and scopes it as its
-own phase.
+## 1. Thesis and principles
+
+**Thesis.** mlx-bun's web chat should not try to be a better ChatGPT. It
+is the local chat surface where the assistant visibly, provably gets
+better at knowing *you* (a git-tracked memory vault you can open and
+diff), where switching to a domain-specific brain is a routing decision
+(LoRA adapters, stackable), and where "local, private, fast" is shown on
+screen (lane badge, tok/s, TTFT, provenance chips) instead of asserted.
+
+**Doctrine (revised 2026-07-06): superset, not wedge.** Concede nothing a
+daily LM Studio / Open WebUI / optiq-Lab user would miss — every
+capability in Appendix A is MATCH, BEAT, or a written SKIP with a
+substrate reason — and stack the differentiators on top. The
+differentiators are the reason to switch, never an excuse for a missing
+table stake.
+
+**Principles.**
+
+1. **Wire before you invent.** The server already had most of the
+   capability (sampler set, structured output, adapters, memory tools,
+   fork, approval gate); the highest-leverage work was wiring, and the
+   remaining open items are still mostly wiring.
+2. **Show, don't assert.** Privacy, speed, memory each get a UI proof.
+3. **Progressive disclosure, one flag.** Chat is the whole app for a
+   first-time user; Quantize / Fine-tune / Build Dataset / Status /
+   Curves / Routes sit behind one persisted Developer toggle.
+4. **Quiet retrieval, legible provenance.** Memory tools fire silently;
+   every consulted article gets a small expandable chip.
+5. **Speed is ambient.** tok/s, TTFT, context-fill, and the serving lane
+   live next to the composer, server-reported, never client-inferred.
+6. **Local-only is an honest simplification.** Single user, own machine:
+   "runs on your machine, with your files" plus a per-call approval gate,
+   not multi-tenant sandbox theater.
+7. **Keyboard-first, mouse-complete.** Every primary action has a key
+   path and a visible one.
+8. **Never hijack.** No spotlight, panel, or approval ever traps focus or
+   blocks the screen; the user is never in a state they have to escape.
+9. **The bar for done is visual polish** at 375 / 768 / 1024 px and
+   desktop, 200 % zoom, `prefers-reduced-motion` honored, no layout shift
+   while streaming. Visual QA at phase boundaries has the standing of
+   tests.
 
 ---
 
-## 2. Where we are
-
-### 2.1 What's genuinely good today
-
-Read from `src/web/app.html` and `src/pi-web.ts` directly, not assumed:
-
-- **A real, working streaming agent bridge.** `PiWebSession` in `src/pi-web.ts`
-  is a legitimately sophisticated per-connection bridge over WebSocket:
-  token/thinking deltas, tool-call cards with live per-call approval gating
-  (`installApprovalGate`, `pi-web.ts:748-774`, 120s auto-deny via
-  `APPROVAL_TIMEOUT_MS`), a follow-up/steering queue distinguishing live
-  mid-stream steering from queued follow-ups (`queue_update` event,
-  `pi-web.ts:390-397`), disk-persisted sessions with new/list/open/fork/delete
-  (`SessionManager.forkFrom`, `pi-web.ts:704-714`), per-turn LoRA injection
-  (`injectAdapter`, `pi-web.ts:412-418`), and per-turn sampling overrides
-  (`injectSampling`) — all through a pure, unit-tested event-mapping layer
-  (`mapEventToFrames`).
-- **A distinctive, cohesive visual language.** A real design-token system
-  (`:root` vars, `app.html:17-33`), gradient hero type, hairline cards,
-  ambient blooms — reads as considered, not a generic Bootstrap clone, across
-  all 7 tabs consistently.
-- **A hand-rolled markdown renderer that is actually careful.** `mdToHtml`/`mdInline`
-  (`app.html:1113-1238`) escape-then-restore code spans, gate link/autolink
-  targets through `mdSafeUrl` against `javascript:` URLs, and support GFM
-  tables/task-lists/fenced code with a working copy button — a rare area
-  where the hand-rolled approach is already competitive with a library.
-- **Real session depth.** Fork/list/delete round-trip through pi's own
-  `SessionManager` and render in the sidebar with per-row actions
-  (`app.html:1735-1736`).
-- **Live perf instrumentation already computed.** Context-window meter
-  (`app.html:1696-1703`) and post-turn tok/s + TTFT (`app.html:1667-1677`)
-  exist — just under-designed as small dim text instead of a first-class
-  signal.
-
-### 2.2 The most damaging gaps
-
-Ranked by severity, from the three-part current-state audit:
-
-1. **(Critical) No personal-memory surface exists at all.** The memory tool
-   suite (`memory_resolve/category/read/section/links/infobox/list/status/search`
-   plus `reference_*`, `src/memory/tools.ts:60-76`) is silently wired into
-   the web chat's tool allowlist whenever a vault exists
-   (`webChatToolAllowlist(memoryEnabled)`, `pi-web.ts:78-83`) and a
-   `memoryHint` is appended to the system prompt — but nothing downstream
-   renders memory tool calls differently from a generic tool card, there is
-   no vault browser, no status indicator, no way to see, search, correct, or
-   trust what the model just read about you. This is the single largest gap
-   between the product's stated identity ("local-to-you", MEMORY.md) and
-   what ships.
-2. **(High) LoRA routing is one flat, undescribed dropdown.** `<select id="chat-adapter">`
-   (`app.html:615-618`) lists raw ids with rank (`${a.id}${a.rank ? " · r"+a.rank : ""}`,
-   unescaped — a live XSS-shaped bug per finding #15 in `web-ui-pass-plan.md`).
-   No description of specialty, no three-state (available/loaded/selected)
-   visibility even though `/v1/adapters/available` already returns
-   `compatible: boolean`, no multi-adapter stacking exposure even though
-   `resolveSpec` in `src/lora.ts` already supports `"a+b"` composition.
-3. **(High) The tool-approval UI is fully built and architecturally dead
-   in web chat.** The approve/deny card with diff preview exists
-   (`app.html:1586-1610`), but the browser session's tool allowlist is
-   hardcoded to `WELCOME_TOOLS = ["read", "web_search"]` plus memory tools —
-   all in `READ_ONLY_TOOLS` — so `bash`/`edit`/`write` (the only
-   `GATED_TOOLS`) never reach the browser surface at all
-   (`buildPiAgentSurface({ webTools: true, codingTools: false })`,
-   `pi-web.ts:543`). The approval dialog only fires in pi's separate
-   terminal/coding-agent embed.
-4. **(High) No message-level actions.** No copy on assistant text (only
-   inside code fences), no edit-and-resend, no regenerate, no branch-from-a-
-   specific-message (only whole-session fork from the sidebar).
-5. **(High) Mobile chat loses the sidebar with no replacement.**
-   `@media (max-width:760px){#chat-sidebar{display:none}}` (`app.html:345`)
-   — no drawer, no hamburger. New chat and all session management become
-   unreachable on a phone viewport.
-6. **(High) No in-chat model switching.** `#nav-model` is a read-only label
-   with no click handler at all (`app.html:527`).
-7. **(Medium) Streaming re-parses the entire message on every token.**
-   `appendDelta` (`app.html:1522-1531`) does `a.text += delta; a.textNode.innerHTML
-   = mdToHtml(a.text)` — full markdown re-parse + innerHTML replace per
-   token. This is O(n²) over a response and becomes the visible bottleneck
-   on exactly the fast, spec-decode-accelerated generations the product
-   exists to showcase.
-8. **(Medium) Sampling popover exposes 3 of ~11 server-supported
-   parameters**, and the "recommended" defaults are hardcoded to one
-   model's shape (`SAMP_REC`, `app.html:1824-1825`, MiniCPM5-tuned),
-   silently wrong for every other served model (`web-ui-pass-plan.md` #14).
-9. **(Medium) No syntax highlighting, no LaTeX, no diagram rendering.**
-   Code fences are plain-escaped monochrome `<pre><code>` (`mdCodeBlock`,
-   `app.html:1139-1143`).
-10. **(Medium/Low) Dark-only theming, ~11 total `aria-*` attributes across
-    2737 lines, no documented keyboard shortcuts beyond Enter/Shift+Enter,
-    no chat search, generic non-personalized hero/empty-state.**
-
-Plus the full tactical list already catalogued in `docs/archive/planning/web-ui-pass-plan.md`
-(queue-bar array-truthiness bug, library staleness, `[object Object]` adapter
-error, dead `status-page.html`, Routes-tab 404 in the compiled binary, stale
-vision copy, etc.) — folded into Phase 0/1 below by number, not repeated in
-full here.
-
-### 2.3 Free wins — backend capability the UI hides today
-
-These require **zero new backend work**, only wiring:
-
-| Capability | Where it lives | Current UI exposure |
-|---|---|---|
-| Full sampler extension set: `min_p`, `xtc_probability`/`xtc_threshold`, repetition/presence/frequency penalties + context windows, `logit_bias` | `src/server.ts` request schema (~`:403-414`, `min_p`/XTC doc-comments; line refs drift with the active dspark branch — grep `min_p` to relocate) | 3 of ~11 fields (temp/top_p/top_k) |
-| `logprobs`/`top_logprobs` | `src/server.ts:425-431, 982-1044` | none |
-| Structured output: `response_format` json_schema/json_object, `guided_grammar`/`guided_regex`/`guided_choice` via xgrammar | `src/server.ts:458-471, 1610-1625` | none |
-| Memory tool suite + `memoryHint` | `src/memory/tools.ts`, `pi-web.ts:78-83` | none — fires silently |
-| `vaultStatus()` (article count, git state, mtimes, schedule state) | `src/memory/vault.ts:1081` | none |
-| `memory_links` (wikilink graph, inbound/outbound) | `src/memory/tools.ts` | none |
-| Adapter three-state model: available (disk) / loaded (mounted) / selected (per-request); `compatible` flag; `"a+b"` stacking | `src/lora.ts`, `GET /v1/adapters/available`, `resolveSpec` | one flat single-select dropdown |
-| Tool-call approval gate with diff/arg preview | `pi-web.ts:748-774`, `app.html:1586-1610` | dead code in web chat (allowlist excludes gated tools) |
-| Fork-from-session (`SessionManager.forkFrom`) | `pi-web.ts:704-714` | sidebar-level only, not per-message |
-| `Responses` API `previous_response_id` resumption, Anthropic `cache_control`/`tool_choice` forcing | `src/responses.ts`, `src/anthropic.ts` | API-only, no chat UI concept |
-| Prompt-cache-hit telemetry (`cache_read_input_tokens`, `cached_tokens`) | server usage payloads | only the aggregate context meter |
-| `GET /fit` hardware-fit prediction (decode tok/s, safe context per SKU) | `src/fit.ts` | none in chat (Status tab only, indirectly) |
-| HLG tone-curve sampling (`hlg` request block) | `src/server.ts:449-457, 483-504`, standalone Curve Designer page | disconnected from main chat composer |
-
-**Two near-free wins that DO need thin server additions** (verified 2026-07-06
-— easy to mistake for pure wiring, budget them explicitly):
-
-- **Lane badge**: the serving lanes (serial / spec-decode / batched) exist
-  internally (`GenerationGateway` routing, `src/server.ts:1162-1168`) but
-  **no field on any usage payload or WS frame reports which lane served a
-  turn**. The perf strip's tok/s / TTFT / context-fill are client-computed
-  already; the lane badge needs a small per-turn `lane` field added
-  server-side (usage payload + the pi-web `turn_end`/usage frame) first.
-- **Memory panel data**: `memory_list`/`memory_search`/`memory_links` are
-  **agent tools** (model-callable, `src/memory/tools.ts`) and `vaultStatus()`
-  is a library function — none have HTTP routes. The vault browser (§5.5)
-  needs thin REST wrappers (e.g. `GET /api/memory/status|list|search|article`
-  + a git log/diff read) added in Phase 2.
+## 2. Architecture as built
+
+### 2.1 The bridge: `src/pi-web.ts`
+
+One `PiWebSession` per WebSocket connection wraps one Pi `AgentSession`.
+Pi events map to frames through the pure, unit-tested `mapEventToFrames`
+(`tests/pi-web.test.ts`). The frame types are exported **type-only** and
+imported by the frontend (`src/web/src/chat.ts`), so contract drift
+between browser and server is a compile error, not a runtime bug.
+
+Client → server (`ClientMessage`):
+
+| Frame | Purpose |
+|---|---|
+| `prompt` (text, images?) | send a turn; images ride as attachments |
+| `abort` | stop the turn (`session.abort()`; verified to abort in-flight tool subprocesses) |
+| `approval` (callId, decision, editedArgs?, alwaysAllow?) | answer a gated tool call; edited args are substituted in place; `alwaysAllow` persists to `src/tool-approvals.ts` |
+| `set_thinking` | reasoning channel on/off (Pi thinking level medium ↔ off) |
+| `set_coding_tools` | opt-in "let it touch files" (default OFF); takes effect on the *next* session because Pi bakes the tool list at session creation |
+| `set_adapter` (id \| null) | active LoRA adapter for subsequent turns (`"a+b"` stacks) |
+| `set_sampling` (scope session \| next_turn, temperature … seed) | full sampler set; `next_turn` is a one-shot override composed over the session-level one |
+| `set_system_prompt` | per-session user prompt layered over the built-in surface prompt (applied per turn via `before_agent_start`) |
+| `new_session` / `list_sessions` / `open_session` / `fork_session` / `delete_session` | Pi session lifecycle |
+| `regenerate` / `edit_resend` / `switch_sibling` | message actions; all are sibling branches in the same session file via `navigateTree` |
+| `context` (AppUiContext) | app-aware assistant: route/view/step + `uiSnapshot`, pushed on route and wizard-step change |
+
+Server → client (`ServerMessage`):
+
+| Frame | Purpose |
+|---|---|
+| `ready` (model, vision, audio, thinking, genDefaults) | capability negotiation; drives image attach, thinking toggle, sampling defaults |
+| `turn_start` / `text_delta` / `thinking_delta` / `turn_end` (lane?) | streaming; `lane` is `"serial" \| "serial+spec" \| "batched"` from `src/serve/lane-registry.ts`, keyed by the assistant message's `responseId`, absent when unknown — never guessed |
+| `tool_start` / `tool_update` / `tool_end` / `tool_approval_request` | tool cards and the approval card |
+| `queue_update` (steering[], followUp[]) | live steering vs queued follow-ups |
+| `history` / `sessions` / `siblings` | transcript replay, sidebar, `< i/n >` toggle data |
+| `context` (tokens, contextWindow, percent) | context-window meter (same name as the client frame, opposite direction) |
+| `coding_tools` (active, pending) | honest current-vs-pending state of the opt-in |
+| `tool_approvals` (alwaysAllow[]) | durable always-allow set, on `ready` and after change |
+| `ui_navigate` / `ui_spotlight` | app-aware assistant side channel, validated server-side |
+| `error` | surfaced in the thread |
+
+Injection hooks (all Pi extension hooks, no custom provider plumbing):
+`before_provider_request` injects `adapter` (`injectAdapter`) and sampling
+(`injectSampling` / `composeSampling`); `before_agent_start` injects the
+user system prompt and the ambient app-context line; `tool_call` /
+`tool_result` carry the approval gate and loop hygiene.
+
+### 2.2 The frontend: `src/web/`
+
+Vanilla TypeScript, no framework. `scripts/build-web.ts` bundles
+`src/web/src/main.ts` into `src/web/app.js`, which `src/server.ts` serves
+at `/assets/app.js`; `app.html` holds the markup and CSS. Runtime stays
+self-contained (no CDN); the only vendored asset is highlight.js
+(`src/web/vendor/`). The bundler errors if a value import ever pulls
+server code into the browser bundle (type-only imports from `pi-web.ts`
+are the intended seam).
+
+| Module | Owns |
+|---|---|
+| `shell.ts` | DOM helpers, toast, focus trap, theme (auto/dark/light), shortcut sheet (Cmd/Ctrl+/), mobile drawer, hash router, connection pill + model identity poll, Developer toggle (`DEV_TABS`), `/dag` probe for the Routes tab |
+| `chat.ts` | the WebSocket, thread render, streaming, tool cards, approval card, queue bar, Sources panel, memory-chip dispatch, app-context push |
+| `composer.ts` | attachments, `#` mention picker, sampling popover (full set + Advanced + scope), system prompt + presets v1 (localStorage), adapter quick-select, perf strip (tok/s, TTFT, context, lane) |
+| `sessions.ts` | sidebar list/search/fork/delete/export, full-text "in messages" search, sibling toggle, per-message actions (copy / regenerate / edit) |
+| `markdown.ts` | `mdToHtml`, block-memoized `renderBlocksIncremental`, highlight pass, `[n]` citation linkify, Canvas toggle for html/svg fences |
+| `memory-panel.ts` | Memory panel, provenance chips, personalized hero chips + first-run consent card |
+| `adapters-panel.ts` | adapter routing table (three states, RAM cost, stacking) |
+| `model-picker.ts` | `#nav-model` popover over `/library` with per-row `/fit` verdict |
+| `hub.ts` | Model Hub panel (`/api/hub/*`), download progress via `/downloads` |
+| `rag.ts` | client-side BM25 over attached-file text |
+| `palette.ts` | command palette (Cmd/Ctrl+K), built via createElement |
+| `assistant.ts` + `ui-catalog.ts` | `captureUiSnapshot`, spotlight resolution/overlay, route + spotlight catalog |
+| `quantize.ts` / `finetune.ts` / `dataset.ts` / `status.ts` | the Developer tabs (job SSE via `api.ts` `jobStream`) |
+| `protocol.ts` / `api.ts` | frontend-only shapes (job stream) and the JSON/SSE helpers |
+
+`sw.js` + `manifest.webmanifest`: shell-only cache-first service worker
+for installability and instant shell — explicitly not offline chat.
+
+### 2.3 Tool surface and trust model
+
+`webChatToolAllowlist(memoryEnabled, codingTools)` in `pi-web.ts`:
+
+- Always: `read`, `web_search` (`WELCOME_TOOLS`) and the three app-aware
+  tools (`get_current_app_context`, `navigate_app`, `spotlight_ui`).
+- With a vault: the memory and reference read tools (`src/memory/tools.ts`).
+- Opt-in only: `grep`, `find`, `ls`, and the gated `bash` / `edit` /
+  `write`.
+
+Everything except the gated three is `READ_ONLY_TOOLS` (auto-allowed).
+Gated calls open the approval card: argument preview, editable arguments,
+allow / deny / always-allow, auto-deny after `APPROVAL_TIMEOUT_MS`.
+Always-allow is durable and keyed by tool name (`src/tool-approvals.ts`).
+Loop hygiene (`LOOP_HYGIENE`): consecutive-identical-call blocking (with
+the prior result inlined, since Pi's `tool_call` hook can only block with
+an error result), a failed-retry budget nudge, and a turn cap with
+force-finish. Format-aware tool-call repair lives in `src/tool-call.ts`
+(`RepairKind`), tagged on the parsed call.
+
+### 2.4 REST the UI depends on
+
+`GET /v1/adapters` (mounted; `ram_bytes`), `GET /v1/adapters/available`
+(on disk; `compatible`, `mounted`), `POST` / `DELETE /v1/adapters`;
+`GET /fit`, `/library`, `/downloads`, `/stats`, `/v1/models`;
+`/api/memory/{status,list,search,article,links,history,diff}` +
+`POST /api/memory/init` (`src/memory/rest.ts`); `/api/hub/{local,search}`,
+`POST /api/hub/download`, `POST /api/hub/serve` (`src/hub-rest.ts`);
+`/api/sessions/{search,export}`; `/api/gc/{plan,execute}`;
+`/api/jobs/:id/stream`. User-facing contract: `docs/reference/server-api.md`.
+
+### 2.5 Regression harness
+
+`tests/web-app.test.ts` (happy-dom) covers the pure client functions:
+streamed-then-finalized output must equal one-shot `mdToHtml` (the
+renderer's own parity oracle), Canvas fence detection and view-state
+survival, `renderQueue` length semantics, `api()` envelope unwrapping,
+`esc()` discipline on every interpolation site, adapter/model/hub row
+renderers, mention picker, BM25, Sources panel, snapshot capture and
+spotlight resolution, catalog validation. `tests/pi-web.test.ts` covers
+the frame mapping, sampling composition, loop hygiene, and app-aware
+tools. Frontend modules type-check in the same `tsc` pass as the server.
 
 ---
 
-## 3. Competitive landscape
-
-| Product | Best-in-class at | Structural ceiling vs. mlx-bun |
-|---|---|---|
-| **mlx-lm** (the oracle) | Correctness/API reference; continuous batching; byte-capped LRU prompt cache; tokenizer-declared capability flags | Zero UI of any kind — its own docs point users elsewhere. No memory, no adapter routing, no structured output, no vision at the server layer. We already win by existing. |
-| **mlx-optiq** | Per-layer sensitivity-guided mixed-precision (weights + KV); Model Arena (blind A/B); Hub (browse+load+chat in one screen); multi-adapter hot-swap via `AdapterActivation` (Python API) | Proved the adapter hot-swap *mechanism* (~50MB/adapter, 10 resident ≈ 6.1GB) but never shipped it above the Python API — no UI, no CLI flag. No memory/personalization concept at all. |
-| **Ollama** | Zero-friction install/run fusion; `ollama launch <tool>` (redirect Claude Code/Codex at itself); `:cloud` suffix hybrid; visible `ollama ps` resource table | No persistent memory, no LoRA routing, thin first-party GUI (borrows its "best chat UI" reputation from third-party clients). |
-| **Open WebUI** | Broadest feature surface (RAG, Channels, RBAC, Functions/Pipelines, MCP, Arena+Elo); native tool calling preserves KV cache | Backend-agnostic by design → zero access to token-level internals (no spec-decode control, no adapter routing, no KV quant visibility). Memory is prompt-injection bounded by context window, not weight-level. Documented long-thread freeze bug (#23067). Enterprise-flavored (RBAC/Channels) — wrong shape for a single-user product. |
-| **LM Studio** | Hardware-fit traffic-light on downloads; User/Power/Developer progressive disclosure; Presets (shareable, model-bound config); transparent inline-vs-RAG; Split View | Closed source. Zero memory/personalization layer — presets are static and manual. Zero LoRA hot-swap UI concept. Electron-heavier than a native shell. |
-| **Odysseus** | Cookbook hardware-aware model recommender; self-evolving Skills (teacher-escalation authoring); persistent Memory (Observer/Reflector, dedup, nightly audit) | Not MLX-native (targets vLLM/llama.cpp, MLX via Ollama's separate backend). Publicly bad security posture (SSRF/auth-bypass CVEs, unsandboxed shell). No adapter/LoRA concept — customization is prompt/Skills-markdown only, not weight-level. "Vibe-coded" reputation. |
-| **AnythingLLM** | Automatic memory extraction (Observer/Reflector, live "3/20" counters, workspace-vs-global scope) — the single feature reportedly driving installs | Memory is workspace-siloed prompt-stuffing, explicitly sent verbatim to whatever provider is configured (no on-device-only guarantee). No LoRA/adapter personalization at all. |
-| **Jan** | Inline non-blocking MCP approval cards + global "Allow All"; router mode (single process, on-demand load/unload); MTP/spec-decode toggle in consumer settings (validates our own dspark serial-routing) | Generic single-thread chat core once you strip the MCP/assistant chrome; no memory, no adapter routing. |
-| **Msty** | Split Chat (parallel synced multi-model panes); Persona Studio + Crew Conversations + Shadow Persona | Closed-source-leaning; feature-sprawl (Turnstiles/Forge Mode naming). No real adapter-weight mechanism — personas are prompt bundles. |
-| **macMLX / mlx-serve** | Genuinely MLX-native, Python-free, same technical identity claim as us (OpenAI+Anthropic API, tiered SSD KV cache, Benchmark tab, quick-launcher hotkey) | Shallow personalization: no cross-session memory, no LoRA routing, basic-to-absent persona support. Proves "MLX-native + fast API" alone is **not** a moat — we already have peers there. |
-| **ChatGPT / Claude.ai** | Artifacts/Canvas; highlight-to-scope-edit; edit-message=branch w/ linear toggle; two-layer transparent memory (explicit facts + background synthesis); Projects (scoped knowledge+memory) | Memory mechanism is a black box even though output is legible — synthesis runs on someone else's infrastructure, unauditable. No tree view (only linear branch toggle) — third-party extensions exist to fill this gap. No LoRA-equivalent — Projects route RAG context, not model weights. Voice mode is a hard, cloud-scale-economics problem. |
-
-**What we'd lose vs. mlx-lm:** nothing — mlx-lm has no UI, no session
-persistence, no adapter routing, no structured output at the server layer,
-no vision at the server layer. Every "loss" is actually a gap we already
-close by existing. The one thing to protect explicitly is **bit-exact
-parity as a mode**: `--batch N` / flag-free paths must keep matching
-mlx-lm's outputs exactly, because "drop-in replacement, losing nothing" is
-a standing product promise (see MEMORY.md), not just a chat-UI concern.
-
-**How we beat Open WebUI / LM Studio (revised):** by conceding nothing and
-then stacking what they can't copy. Concretely, per the
-[beat matrix](web-chat-beat-matrix.md): we MATCH every table stake a user
-would miss — system prompts/presets, document chat with citations, in-app
-model browse/download (optiq's Hub flow, plus a real `/fit` verdict per row
-that optiq lacks), sandboxed HTML canvas, per-message sampling, session
-auto-titling, export, shortcuts, themes, mobile — and we skip only where the
-engine lacks the substrate (image generation, voice), in writing. The two
-exceptions we deliberately don't chase: Open WebUI's multi-tenant enterprise
-plumbing (Channels, RBAC) — wrong shape for a single-user local product, a
-scope decision rather than a gap — and cloud-scale voice. THEN the wedge:
-we're the only surface whose memory is **weight-level-capable** (git-tracked
-vault + nightly synthesis today, memory-LoRA rung on the roadmap — not a
-context-window fact list with a hard cap) and whose adapter routing is
-**real** (LM Studio has zero adapter concept; Open WebUI is
-backend-agnostic and structurally cannot see adapters at all). Both are
-already true at the engine level — the redesign's job is to make them
-visible, legible, and trustworthy, which neither competitor's architecture
-permits even if they wanted to.
-
-*(Caveat: the Odysseus row above rests on the research agent's public-web
-identification — Cookbook recommender / Skills / Observer-Reflector memory /
-SSRF CVEs. If that isn't the Odysseus meant, that row needs a re-run;
-nothing else in this doc depends on it.)*
+## 3. Surfaces
+
+### 3.1 Onboarding / hero
+
+Server-driven chips instead of static marketing copy: one drawn from real
+vault content when `/api/memory/status` reports articles, one offering an
+available adapter, and a first-run, skippable memory-consent card that
+creates the vault via `POST /api/memory/init` — consent happens in the
+same first minute as the first chat, not as a CLI step discovered later.
+The system prompt (`buildWebChatSystemPrompt`) is deliberately short: a
+long welcome prompt made the ~1B starter model greet instead of answer.
+
+### 3.2 Composer and message actions
+
+- Sampling popover: temperature / top_p / top_k plus an Advanced
+  disclosure with `min_p`, XTC, repetition / presence / frequency
+  penalties with context size, and seed; defaults from `ready.genDefaults`
+  (per-model, not a hardcoded shape); scope control session vs next turn.
+- System prompt field + presets v1 (name + prompt + sampling; localStorage
+  only, no wire shape).
+- Adapter quick-select (`#chat-adapter`) with a gear to the routing table
+  (§3.6). A swap note says so: a new adapter starts a fresh KV segment.
+- `#` mention picker: attached files and vault articles as one retrieval
+  gesture.
+- Perf strip: tok/s, TTFT, context-fill, lane badge (from `turn_end.lane`).
+- Per assistant turn: copy, regenerate; per user turn: edit-and-resend
+  with a `< i/n >` sibling toggle. All three are branches of the same Pi
+  session (no new file), so history stays in one place. A full tree
+  *view* is deliberately not built (open, §5).
+
+### 3.3 Streaming render
+
+`renderBlocksIncremental` splits the message at block boundaries, leaves
+settled blocks alone, and re-parses only the live tail, throttled to
+animation frames; `finishStreaming` runs one full `mdToHtml` pass so the
+final DOM is byte-identical to a one-shot render (tested). Syntax
+highlighting (vendored highlight.js) applies to completed blocks only.
+Steering and follow-up queues render from `queue_update` with length
+checks (the array-truthiness bug is fixed and regression-tested).
+
+### 3.4 Tool cards and approval
+
+Generic tool calls keep the running / ok / fail card. Memory and reference
+tool calls render as a **provenance chip** (citation style, expandable,
+opens the article in the Memory panel) — the "assistant looked something
+up about you" moment reads differently from "ran a shell command". The
+approval card (§2.3) is inline and non-blocking. The settings copy says
+that toggling coding tools applies to the next new chat (`coding_tools`
+frame), never implying immediacy Pi can't deliver.
+
+### 3.5 Memory panel
+
+Reachable from the chat sidebar, not a nav tab (memory is chat-adjacent,
+it is the product, not a developer tool). Two screens: list (status strip
+from `vaultStatus()`, search, personal articles and a `Reference/`
+sub-list) and article (rendered content, inbound/outbound wikilinks line,
+a History toggle backed by `/api/memory/history` and `/api/memory/diff` —
+the "watch it self-heal after a correction" moment made literal, which
+cloud memory UX structurally cannot offer). The agent-tool surface stays
+read-only; only nightly synthesis writes (`docs/design/dreaming-nightly-pipeline.md`,
+`docs/design/dreaming-nightly-pipeline.md`). The plan's wikilink *graph* view is not built —
+the panel shows a links line (open, §5).
+
+### 3.6 Model and adapter switching
+
+**Model picker.** `#nav-model` opens a popover over `/library`, each row
+with the `/fit` verdict computed for this Mac (predicted decode tok/s,
+safe context). There is **no in-process model swap**: the runtime is one
+model per process (like mlx-lm), `POST /api/hub/serve` answers
+`restart_required` with the exact `mlx-bun serve` command, and the picker
+offers that command to copy. The live-swap mechanism that exists
+(`ModelPool` spawn-overlap) lives only on the opt-in `--isolate` proxy,
+which 501s `/ws/chat` — see `docs/reference/server-config.md`. Model Hub (`hub.ts`)
+adds browse / HF search / download with progress; it never auto-downloads
+or auto-serves.
+
+**Adapters — the three-state model** (folded from
+`docs/design/web-chat-redesign.md`; the design keeps the states separate and lets
+the user move an adapter between them explicitly):
+
+1. **Available** — on disk, compatible or not, costs nothing.
+   `GET /v1/adapters/available` scans `~/.cache/mlx-bun-finetunes/` and
+   `~/.cache/mlx-bun/adapters/` (`listAvailableAdapters`, `src/lora.ts`);
+   `id` = directory basename; `rank` / `scale` / `base_model` read from
+   `optiq_lora_config.json` or `adapter_config.json`, with the base repo id
+   recovered from the stored snapshot path.
+2. **Loaded / mounted** — `POST /v1/adapters {id, path}` into the running
+   server; costs memory (`ram_bytes` on `GET /v1/adapters`);
+   `DELETE /v1/adapters/:id` unmounts.
+3. **Selected for this request** — the `adapter` field: `"none"` (default)
+   | `"<id>"` | `"a+b"` (stacked, `resolveSpec`). Unknown id throws —
+   loud, never a silent base fallback.
+
+The routing table (`adapters-panel.ts`, behind the composer gear) shows
+every on-disk adapter with badges for all three states, base-model / rank
+/ size / RAM meta, mount / unmount / select / clear actions, and "stack"
+tick-boxes that compose two mounted adapters as `"a+b"`. Selecting an
+unmounted adapter mounts it first, then sends `set_adapter`; the
+`before_provider_request` hook injects the id. Invariants: default none;
+an adapter must be mounted before it can be selected; the quick-select and
+the table are two views of one selection.
+
+Compatibility as built: `compatible = base_model == null ||
+bareName(base_model) === bareName(servedModelId)` — a bare-name match on
+the base model, not the repo-id match the original note described.
+
+The **CLI twin** is the Pi extension
+`scripts/packaging/pi-extensions/mlx-bun-adapter.ts` (`/adapter`,
+`/adapter <id>`, `/adapter off`; server via `MLX_BUN_URL`; same
+`before_provider_request` mechanism, in-memory selection). It belongs to
+the CLI reference (`docs/reference/cli.md`, alongside `serve --adapter`),
+not to this doc.
+
+### 3.7 Sessions
+
+Sidebar with title filter, a full-text "in messages" fallback
+(`/api/sessions/search`, confined to the session dir), fork / delete /
+export (Markdown / JSON via `/api/sessions/export`). Titles derive from
+the session name or first user message (`toSessionListItems`). On narrow
+viewports the sidebar becomes a focus-trapped slide-over drawer.
+
+### 3.8 Information architecture and settings
+
+Chat is the product; everything else is a developer tool. One persisted
+Developer toggle (`shell.ts`) reveals Quantize / Fine-tune / Build Dataset
+/ Status / Curves / Routes. Curves is a separate full page (`/curves`,
+labelled as such); Routes is a hash route whose tab visibility is gated by
+a `/dag` probe so the compiled binary never shows a broken iframe. Deep
+links keep working regardless of the toggle. Status's RAM/KV "Memory"
+card keeps meaning system memory; the vault is always "Memory panel".
+Settings group as Chat (sampling, thinking, coding-tools opt-in,
+always-allow list), Memory (status, consent), Developer (the toggle).
+
+### 3.9 Chat-with-files (RAG v1)
+
+Attached text files below an inline threshold go into the prompt verbatim;
+above it, a pure client-side BM25 retriever (`rag.ts`, no embeddings, no
+network) selects top chunks, injects them as a numbered context block, and
+the reply's `[n]` markers linkify into a collapsed Sources panel.
+Citations survive regenerate / edit; they are not persisted into history
+replay (needs a `HistoryItem` field — open). A vector upgrade over
+`src/embed.ts` is explicitly later.
+
+### 3.10 Code and Canvas
+
+Fenced code gets copy + highlighting. `html` / `svg` fences get a
+Preview | Source toggle: a lazily created `sandbox="allow-scripts"`
+`srcdoc` iframe, never `allow-same-origin`, height-clamped, view state
+preserved across re-render and streaming finish. The persistent Artifacts
+side panel is not built (open, §5).
+
+### 3.11 The app-aware assistant
+
+The agent can see and act on the app it lives in — no screenshots, no
+vision model in the loop. Product theory: guided tours fail on timing
+(help before intent), chatbot help fails on grounding (intent without
+state); this is the only shape with both, and the acceptance test is a
+feeling — someone is there with you.
+
+- **See:** on route / wizard-step change the browser pushes a `context`
+  frame (route, view, step, `uiSnapshot` of visible interactive elements,
+  capped ~120, agent chrome excluded). `get_current_app_context` returns
+  it; a compact ambient line (route / view / step) is auto-prepended to
+  every turn so the model never answers blind.
+- **Navigate:** `navigate_app` validated server-side against the route
+  catalog → `ui_navigate` → the hash router. Reversible, no approval.
+- **Point:** `spotlight_ui` resolves ref → selector → label → catalog
+  target against the last snapshot and shows a hand-rolled overlay:
+  `pointer-events: none` (the control stays clickable through it), no
+  focus trap, auto-dismiss in seconds, any input dismisses, reduced-motion
+  honored. Model-invented selectors are a miss, never a throw.
+
+Scope guard: these tools exist only on the pi-web bridge (never the
+OpenAI / Anthropic API surface); the snapshot never includes cross-origin
+content; approval-gated `ui_act` (fill this form, start this job) is open
+work and must render through the §3.4 approval card.
+
+### 3.12 Speed and trust as ambient proof
+
+The perf strip's lane badge is server-reported per turn (risk: a stale or
+wrong badge is a P0, because the differentiator rests on it being
+trustworthy). Context-fill comes from Pi's `getContextUsage`. What is still
+missing from the trust surface is listed in §5.
 
 ---
 
-## 4. Design principles
-
-1. **Concede nothing; differentiate on top.** The beat matrix
-   ([web-chat-beat-matrix.md](web-chat-beat-matrix.md)) is the coverage
-   contract: every capability row is MATCHed, BEATen, or SKIPped in writing
-   with a substrate reason — never silently absent. Within that, the
-   memory/LoRA flywheel gets the design attention (it's the reason to
-   switch), but "it's just table stakes" is never a reason a table stake
-   ships worse than the best competitor's version. The test for every
-   release: would a daily LM Studio / Open WebUI / optiq-Lab user miss
-   anything? If yes, that gap has a name and a written decision.
-2. **Show, don't assert.** Privacy, speed, and memory are all *provable*
-   locally (no network egress, real tok/s, an inspectable git-tracked vault).
-   Every one of these claims gets a UI element that proves it, not prose
-   that states it. This mirrors the CLAUDE.md ground rule "every perf claim
-   gets a number" — extend it to every trust claim getting a UI proof.
-3. **Progressive disclosure, one flag, not two apps.** LM Studio's
-   User/Power-User/Developer toggle is the reference: one codebase, one
-   settings switch, not a fork. Quantize/Fine-tune/Build-Dataset/Curves/Routes
-   collapse behind a single "Developer" toggle so Chat is the whole app for
-   a first-time user.
-4. **Quiet retrieval, legible provenance.** Memory tools fire silently by
-   design (never a context dump) but every consulted source gets a small,
-   dismissable, expandable citation — the AnythingLLM/ChatGPT memory-panel
-   legibility bar, applied to a strictly deeper (weight-level) mechanism.
-5. **Speed is ambient, not a stat you have to look up.** tok/s, TTFT,
-   context-fill, and which serving lane is active (serial / spec-decode /
-   batched ×N) are always visible near the composer, the way LM Studio 0.4
-   made context-fullness and processing state ambient chrome.
-6. **Wire before you invent.** Prefer surfacing an existing backend
-   capability over designing a new one. Section 2.3's free-wins table is the
-   backlog triage order, not a suggestion.
-7. **Local-only is an honest simplification, not a hedge.** Single-user,
-   local-machine execution means tool access, file writes, and shell
-   commands can be presented plainly ("runs on your machine, with your
-   files") instead of building sandbox theater Open WebUI/Odysseus need for
-   multi-tenant deployments we don't have.
-8. **Keyboard-first for the power user, mouse-complete for everyone else.**
-   Every primary action (new chat, send, stop, regenerate, switch
-   model/adapter, command palette) has a keyboard path, but nothing is
-   keyboard-only-discoverable — mirror ChatGPT's documented shortcut sheet
-   (`Cmd/Ctrl+/`) as the baseline, not a stretch goal.
-9. **The bar for done is perfect visual polish.** No overlapping text at
-   any viewport width, no animation jitter, no layout shift on stream, no
-   flash of unstyled/mispositioned anything. A feature that works but
-   looks unfinished is not done — it ships when it feels like a polished,
-   professional, excellent experience. Concretely: every animation is
-   compositor-friendly (transform/opacity only), interruptible, and
-   respects `prefers-reduced-motion`; every overlay/popover is tested at
-   375px, 768px, 1024px (iPad — where the 2026-07-06 nav-clip /
-   placeholder-wrap / pill-overflow jank actually surfaced), and desktop
-   widths and at 200% zoom; streaming text never
-   causes scroll jank or reflow of completed content (§5.3's memoization
-   is a polish requirement, not just a perf fix). Visual QA at phase
-   boundaries is part of the definition of done, same standing as tests.
+## 4. Tech-stack decisions
+
+- **Vanilla, no SPA framework.** The framework-native ideas (block
+  memoization, windowing, focus traps) port as functions; a migration is a
+  multi-month distraction at this file's complexity. Recorded so a future
+  session doesn't relitigate it without new evidence.
+- **Typed module split over a single inline script.** The split's payoff
+  is the typed WS/API contract, not navigability: the Phase 0 bug class
+  (field-name drift, array truthiness, response-shape mismatch) became
+  build errors.
+- **Vendored, no-CDN assets only.** highlight.js is vendored; KaTeX was
+  planned and is not vendored (open).
+- **PWA = installability + instant shell**, never offline chat.
+- **Theming:** auto / dark / light via `data-theme`, follows
+  `prefers-color-scheme`; animations transform/opacity only and gated by
+  `prefers-reduced-motion`.
+- **A11y:** `role="log" aria-live="polite"` on the thread, focus traps with
+  Escape-returns-focus on every overlay, a shortcut sheet with ChatGPT's
+  vocabulary where sensible.
+- **Hard rule for every overlay:** the spotlight is not a modal and must
+  never be wired into the modal machinery (`closeTopOverlay` / `trapFocus`).
 
 ---
 
-## 5. The redesigned chat experience
-
-### 5.1 Onboarding / empty state
-
-**New vs. today:** the hero ("Local. Private. Yours.", `app.html:564-570`)
-currently shows 4 static marketing chips regardless of state. Replace with
-server-driven chips: if a vault exists, one chip is drawn from real content
-("Ask about {last-synthesized-article}"); if adapters are available, one
-offers to try one; if this is a first run, a one-line skippable memory
-consent card appears inline ("Let mlx-bun learn from this conversation? Two
-clicks, reversible.") — because `mlx-bun memory init` today is CLI-only and
-invisible to anyone who never opens a terminal. This directly serves the
-"install to chat under a minute" north star: personalization consent should
-happen in the same 60 seconds as first chat, not as a separate discovered
-CLI step.
-
-**Backend used:** `vaultStatus()` (`src/memory/vault.ts:1081`), `GET
-/v1/adapters/available`, existing hero markup as scaffold.
-
-### 5.2 The composer and message actions
-
-**Composer additions (extend, don't rebuild):**
-- **LoRA/persona chip** replacing the bare `<select id="chat-adapter">`:
-  active adapter's name + one-line specialty description (falls back to
-  "general-purpose" until metadata exists), available/incompatible adapters
-  grayed by the `compatible` flag `/v1/adapters/available` already returns.
-  A note appears on swap ("starts a fresh KV segment") since that's true.
-- **Sampling popover** grows from 3 sliders to the full server-supported set
-  behind an "Advanced" disclosure: `min_p`, XTC (`xtc_probability`/`xtc_threshold`),
-  the three penalty families with context windows, `logit_bias`, seed.
-  Defaults resolve per-model from `generation_config.json` (fixing the
-  hardcoded MiniCPM5-shaped `SAMP_REC`, `web-ui-pass-plan.md` #14) instead of
-  one hardcoded shape.
-- **Perf strip** under the composer, always visible: tok/s, TTFT,
-  context-fill %, and a **lane badge** (`serial · spec-decode active` /
-  `batched ×N` / `parity mode`) — replacing the small dim tok/s text. This is
-  the one thing no competitor's UI has, because none of them control the
-  serving lane the way we do.
-- **Unified `#`/`@` retrieval mention**: typing `#` opens a picker across
-  *both* attached files and vault articles as one retrieval gesture (Open
-  WebUI's `#`-mention pattern, extended so "attach a file" and "search my
-  memory" are the same interaction, not two parallel systems).
-
-**Message actions (new — currently only code-block copy exists):**
-- Per assistant turn: copy full message, regenerate, "branch from here"
-  (reuses `SessionManager.forkFrom`, currently only exposed at the
-  whole-session level).
-- Editing a user message creates a lightweight sibling with a `< 1/2 >`
-  toggle above it — the ChatGPT/Claude convention users already have muscle
-  memory for. **Do not** build a full tree-view graph in this phase (real
-  effort neither incumbent bothered to ship); a tree view is a Phase 5
-  stretch differentiator (see §6, and note both cloud incumbents ship only
-  the linear toggle — a from-scratch tree view is a genuine, buildable gap
-  we can claim later, not a launch requirement).
-
-### 5.3 Streaming + steering
-
-Fix the O(n²) re-render (`appendDelta`, `app.html:1522-1531`) by splitting
-into completed-block + live-tail-block: parse paragraph/fence boundaries,
-skip re-parsing completed blocks, only reprocess the tail (Streamdown/t3.chat's
-block-memoization pattern, ported as an idea into the existing vanilla
-renderer — no framework needed). Ship self-hosted syntax highlighting (Shiki
-or a lighter alternative, applied only to completed blocks) and KaTeX for
-math in the same pass, since code fences currently render as monochrome
-plain text. This is not optional polish: on a fast local model (the whole
-product pitch), the current renderer makes the *frontend* the visible
-bottleneck.
-
-Steering and the follow-up queue already work server-side
-(`queue_update` carries distinct `steering`/`followUp` arrays,
-`pi-web.ts:390-397`) but render broken today (truthy-tested instead of
-length-tested arrays, `web-ui-pass-plan.md` #2) — fix as part of this pass
-since it's touching the same render path.
-
-### 5.4 Tool-call & approval display
-
-Keep the existing running/ok/fail visual language
-(`app.html:1545-1584`). Two concrete changes:
-1. **Wire the web chat's tool allowlist to include gated tools** (`bash`/`edit`/`write`)
-   behind an explicit, off-by-default "let it touch files" toggle in
-   settings (`buildPiAgentSurface({ codingTools: true })` on that path) —
-   so the already-built approval-card UI (`app.html:1586-1610`) stops being
-   dead code. Extend the card to show editable arguments before
-   approve/deny (LM Studio's pattern, beyond mlx-bun's current binary
-   allow/deny) and persist "always allow this tool" to a real config file
-   keyed by tool identity (fixing the fragile-persistence failure mode
-   documented against Claude Desktop/Cowork in the research).
-2. **Give memory tool calls a distinct card style** — not the wrench icon,
-   something closer to a citation chip — so "the assistant looked something
-   up about you" reads differently from "the assistant ran a shell
-   command." This is the single highest-leverage rendering change in the
-   whole plan given §2.2 finding #1.
-
-### 5.5 Memory surface (view / edit / forget, LoRA indicator)
-
-A real **Memory panel**, reachable from the chat sidebar (not a 7th peer tab
-to Quantize/Fine-tune — memory is chat-adjacent, one click away, because
-it's the product, not a dev tool). Contents:
-- **Vault browser + search**, backed by `memory_list`/`memory_search`,
-  with a `Reference/` sub-view for mlx-bun's own docs (clearly separated
-  from personal articles) via `reference_list`/`reference_search`.
-- **Per-article view with a git-log/diff toggle** — the "watch it self-heal"
-  moment made literal: after a correction, a user can see the exact article
-  diff from the next nightly synthesis. This is the sharpest wedge against
-  ChatGPT/Claude's memory UX, which shows *what* is remembered but never
-  *how* — because their synthesis runs on someone else's infrastructure.
-  Ours runs locally and the mechanism itself is inspectable.
-- **Status strip**: `vaultStatus()` already returns article count, git
-  state, last-synthesis time — render it plainly.
-- **Wikilink graph view** using `memory_links` (inbound/outbound per
-  article) — cheap given the tool already resolves this data; a genuine
-  "knowledge graph of what the model knows about you" view nobody else in
-  the survey has because nobody else has a git-tracked personal wiki.
-- **Per-turn provenance chip** on any assistant message that consulted
-  memory tools (from §5.4) — links into this same panel.
-
-### 5.6 Model & adapter switching
-
-Make the nav model label (`#nav-model`, currently a dead `<span>` with no
-click handler) clickable to a lightweight picker: what's loaded, hardware-fit
-signal (green/yellow/red, LM Studio's pattern) computed against this
-specific Mac's measured bandwidth via `GET /fit` — not a generic heuristic,
-since we already have per-SKU fit prediction in `src/fit.ts`. Model swap
-should preserve session history and re-prime cleanly (every competitor
-surveyed is documented as bad at this — LM Studio has an open bug where
-switching models clears the selector and forces manual reselection; a
-genuinely smooth swap is a believable, demonstrable win).
-
-Adapter selection: promote from flat dropdown (§5.2) to a routing table in
-settings showing all three states (available on disk / loaded+mounted /
-selected for this request) with RAM cost visible per loaded adapter (optiq's
-own numbers: ~50MB/adapter, 10 resident ≈ 6.1GB — a concrete budget to design
-the UI affordance against), and expose `"a+b"` stacking since `resolveSpec`
-already supports it server-side.
-
-### 5.7 Sessions / projects
-
-Add a search box above the session list (trivial, local-only, full-text —
-no reason to skip it given Claude's own title-only search is a widely-cited
-annoyance and we have no server-cost constraint). On mobile, replace the
-currently-vanishing sidebar (`app.html:345`) with a slide-over drawer
-triggered by a hamburger — nothing fancier needed for v1.
-
-A lightweight **persona/bundle** concept (system prompt + adapter selection
-+ tool allowlist + memory scope, saved and nameable) is a Phase 5 item (see
-§6) — not required for the core redesign, but the composer/settings data
-model should be shaped so a "save this as a named bundle" action is a
-thin layer on top, not a re-architecture.
-
-### 5.8 Settings
-
-Collapse into three groups: **Chat** (sampling, thinking toggle, tool
-allowlist/gating), **Memory** (vault path, consent, synthesis schedule
-status — read from `memory_status`), **Developer** (the progressive-disclosure
-gate for Quantize/Fine-tune/Build-Dataset/Curves/Routes — see §8).
-
-### 5.9 Structured output
-
-Server already supports `response_format` json_schema/json_object and
-`guided_grammar`/`guided_regex`/`guided_choice` via xgrammar
-(`src/server.ts:458-471`) with zero UI anywhere. Ship a lightweight
-schema/grammar picker in the Advanced sampling disclosure (§5.2) — even a
-raw JSON-schema textarea with live validation is pure whitespace relative to
-every competitor (mlx-lm has no `response_format` at all; most local UIs
-don't expose it even when their server supports it).
-
-### 5.10 Vision / multimodal
-
-Vision is already wired end-to-end correctly (capability negotiation via
-the `ready` frame's `vision` flag, `pi-provider.ts:93-100`). The only fix
-needed here is the stale copy bug (`web-ui-pass-plan.md` #6: "switch to the
-12B for vision" is wrong now that e2b/e4b/26B/31B are all vision-capable via
-`visionCapable(m)`). No new UI work required for v1 beyond that correction
-and folding image/doc attach into the unified `#` retrieval gesture (§5.2).
-
-### 5.11 Artifacts / code
-
-Fenced-code rendering gets a copy button (already exists) plus syntax
-highlighting (§5.3). **Canvas v1 — fenced HTML rendered in a sandboxed
-iframe with a view-source toggle — is a Phase 3 item** (optiq Lab ships
-exactly this today; "I miss optiq's Canvas" is a disallowed outcome). The
-full Artifacts/Canvas *side-panel* (persistent, versioned,
-highlight-to-scope-edit) stays a Phase 5 stretch — both cloud incumbents'
-own experience (OpenAI partially walked Canvas back in May 2026 for the
-common case) argues for opt-in-by-heuristic (auto-open above ~10 lines of
-code / high fence density), not a mandatory pane.
+## 5. Open work
+
+Nothing below has a PLAN.md phase; it is the backlog this doc owns.
+Ordered by leverage.
+
+**Residuals from the shipped phases (verified against src 2026-08-23):**
+
+- KaTeX / math rendering — planned in the render pass, not vendored.
+- Structured-output picker (JSON schema / grammar textarea with live
+  validation) — server supports `response_format` and
+  `guided_grammar|regex|choice`; no composer UI.
+- Wikilink graph view — `/api/memory/links` exists; the panel renders a
+  links line only.
+- Session-preserving live model swap — blocked on a reload seam at the
+  engine boundary (`docs/reference/server-config.md`); until then the Hub / picker
+  hand back the restart command.
+- RAG citations persisted into history replay (`HistoryItem` field).
+- Long-thread windowing (anchored-to-bottom DOM recycling) — not built;
+  months-long sessions are more likely here than in a stateless chatbot.
+- Adapter metadata: the discovery rows are indistinguishable when several
+  adapters share base / rank / scale; the trainer should stamp task / data
+  / date (and a stable repo id) into the adapter config at save time.
+- Overnight training from the web UI ("start before bed, wake up to an
+  adapter"): the fine-tune wizard streams a job (`/api/jobs/:id/stream`),
+  but a reconnectable monitor with completion → "available" hand-off is
+  not designed. `--force-adapter` (serve pinned to one adapter for an
+  unattended pipeline) is a CLI item, tracked in the CLI reference.
+
+**Trust and speed differentiators (former Phase 4):**
+
+- "Temporary chat" pill — provably never written to disk.
+- Per-session disk-touch tally (files read / written) as a trust log
+  distinct from tool cards; per-session "zero cloud calls" indicator.
+- Prompt-cache-hit telemetry per turn ("reused N cached tokens").
+- logprobs-derived confidence overlay (server already returns
+  `logprobs` / `top_logprobs`).
+- HLG tone-curve controls in the composer (today only on the Curves page).
+- App-aware assistant maturity: a persistent minimized affordance on every
+  view (Developer tabs included) that minimizes on engagement and
+  re-expands only on invocation; process-state enrichment of the snapshot
+  (job progress, field completion) beyond route / step.
+
+**Stretch (former Phase 5; pick one or two at a time):**
+
+- Full conversation tree view over the sibling groups Pi already stores.
+- Persona bundles (system prompt + adapter + tool allowlist + memory scope)
+  — presets v1's data model is shaped so this is a thin layer.
+- Compare / split view: one prompt against base vs adapter, or two quant
+  tiers, side by side with per-pane tok/s (optiq Arena's mechanics, plus
+  the comparisons Arena cannot express).
+- Approval-gated `ui_act` tools.
+- Artifacts side panel, opt-in by heuristic; Mermaid; vector RAG upgrade;
+  MCP client; Skills-style procedural memory (all Appendix A rows with a
+  written state).
+- Memory that graduates into a trained memory LoRA — roadmap; the UI must
+  never claim it before it ships.
+
+**Risks that stay live:** the superset doctrine ballooning scope (keep
+items independently shippable; re-audit Appendix A at each boundary, not
+mid-stream); the lane badge going stale in a mode transition (server-
+driven only); always-allow persistence regressing to the fragility seen in
+other desktop agents (versioned config keyed by tool identity, migration
+test); the memory diff view unsettling a user (user-initiated inspection,
+pause / reset one click away).
 
 ---
 
-## 6. Differentiator features (deep treatment)
-
-These five are the headline bets — the features that make this unmistakably
-mlx-bun and structurally hard for any surveyed competitor to copy without
-rebuilding their own serving stack or personalization architecture.
-
-### 6.1 Memory as a visible, weight-grounded flywheel
-
-Not a fact list (AnythingLLM, ChatGPT, Claude) — a git-tracked vault where
-chat-time reads are strictly read-only (`src/memory/tools.ts`) and only
-nightly synthesis writes (`docs/design/memory-system.md`). The UI's job:
-(a) a provenance chip on any turn that consulted the vault, (b) a real
-Memory panel (§5.5) with a git-diff toggle so a user can *watch* an article
-evolve after a correction — literally the "self-healing through ordinary
-use" story (`docs/design/the-dreaming-master-plan.md`) made visible instead
-of asserted, (c) eventually, a visible marker distinguishing "this fact
-lives only as raw context" vs. "this has been baked into your memory LoRA" —
-a distinction no competitor's architecture can even express, since none of
-them synthesize memory into trained weights.
-
-### 6.2 LoRA-per-context routing, made legible
-
-optiq proved the resident-multi-adapter mechanism works (~50MB/adapter,
-6.1GB for 10 resident vs. 56GB for 10 full model copies) but shipped it
-Python-API-only. LM Studio's own bug tracker confirms adapter hot-swap is
-unsupported. mlx-bun already has `/v1/adapters`, adapter-namespaced KV, and
-per-request injection working end to end (`src/lora.ts`,
-`docs/design/adapters-end-to-end.md`). The UI work is turning this into a
-routing table: which adapter is active per-surface, why (matched from
-surface / explicitly selected), its RAM cost, and — the genuine escalation —
-support for stacking (`"a+b"`) directly in the picker, something the backend
-already does and no UI anywhere exposes.
-
-### 6.3 On-device privacy, proven not asserted
-
-A one-click "Temporary chat" pill (matching ChatGPT/Claude's composer-header
-convention) that — unlike cloud temporary chat, which OpenAI's own
-disclosures say is still retained 30 days for "safety" — can be made
-**provably** never written to disk at all, because we control the whole
-stack. Pair this with a per-session "zero cloud calls" indicator and a
-running tally of what the agent touched on disk (files read/written this
-session) as a trust surface distinct from generic tool-call cards.
-
-### 6.4 Speed surfaced as a felt experience, not a hidden stat
-
-The perf strip (§5.2) — tok/s, TTFT, context-fill, lane badge — is the
-direct answer to "serving speed is the user metric" (MEMORY.md). No
-surveyed competitor (including the MLX-native peers macMLX/mlx-serve, which
-do have a Benchmark tab) integrates lane-awareness (serial vs. spec-decode
-vs. batched) into the *live chat surface* itself — they show a benchmark
-number in a separate tab, we can show which lane is running the actual
-conversation right now. This directly explains latency variance instead of
-letting it read as jank, and ties concretely into the active `dspark`
-spec-decode branch.
-
-### 6.5 Local tool transparency as a trust surface, not a permission dialog
-
-Every competitor's tool-approval UX is either a blocking modal (old Jan,
-Claude Desktop) or a binary allow/deny with no argument preview (most).
-mlx-bun already has the approval round-trip built (`pi-web.ts:748-774`); the
-differentiator is finishing it properly: inline non-blocking cards (Jan
-v0.8.0's pattern) with editable arguments before execution (LM Studio's
-pattern) and durable, tool-identity-keyed "always allow" persistence (fixing
-the failure mode publicly documented against Claude Desktop). Combined with
-§6.3's disk-touch tally, this reframes "the agent can run bash" from a scary
-permission dialog into a legible, reviewable log — the natural trust story
-for a single-user local product that structurally doesn't need Open WebUI's
-sandbox theater.
-
-### 6.6 The app-aware assistant (added 2026-07-06)
-
-The chat agent can see and act on the app it lives in. Verified against the
-full beat matrix: **no product in the surveyed field has this** (matrix
-Axis 12). Reference implementation proven in-house on the same pi SDK:
-`~/Code/PortfolioManager`.
-
-**Why this matters (the product theory, not the feature):** both existing
-in-app help paradigms fail on the same axis from opposite ends. Guided
-tours fail on *timing* — they front-load help before intent exists, so
-users skip them and the knowledge is gone by the time they care. Chatbot
-help fails on *grounding* — it arrives at the moment of intent but blind:
-no idea what screen you're on, what step you're in, what you just tried; a
-search bar with a face, and users feel it immediately. The app-aware
-assistant is the only shape with both: help at the moment of intent,
-grounded in current state. Instead of screenshotting and circling a thing
-for an AI, you ask — and it highlights, navigates, takes you there. **The
-acceptance test is a feeling: like someone is there with you.** Not a
-tour, not a search bar — presence.
-
-**Design consequences the feeling forces** (a naive port misses these):
-
-- **Ambient across every view, not confined to the Chat tab.** A guide you
-  must leave the Quantize tab to talk to is blind again by construction.
-  Ship a lightweight persistent assistant affordance on every view —
-  Developer tabs included — with the Chat tab as the deep surface
-  (PortfolioManager's floating-panel pattern proves it on our SDK).
-  **And the panel itself is never intrusive:** the moment the user engages
-  the page (clicks Quantize, focuses a form field), the panel minimizes
-  with a smooth animation to a small persistent affordance — you know it's
-  there, it isn't blocking you. It re-expands on invocation, never on its
-  own initiative.
-- **The snapshot carries *process state*, not just elements.** "Knows what
-  part of the process you're in" means the context push includes wizard
-  step (Quantize step 2 of 4), running job progress, which fields are
-  filled — beyond PortfolioManager's visible-element list.
-- **Never answer blind.** Tool-pull alone (the agent *choosing* to call
-  `get_current_app_context`) sometimes won't fire, and one ungrounded
-  answer breaks the "with you" illusion. Auto-attach a compact ambient
-  context line (route + view + step) to every turn; the deep snapshot
-  stays on-demand.
-- **Choreography:** the spotlight lands *while* the explanation streams
-  ("it's right here —" and the highlight appears), not as a report
-  afterwards.
-- **Never hijack (hard constraint).** The worst tour UX is blocking the
-  screen when the user wants to DO something. The spotlight is an overlay,
-  never a modal: it traps no focus, blocks no clicks (the highlighted
-  control stays clickable *through* it), auto-dismisses in seconds, and
-  any user input dismisses it instantly. Navigation only happens when the
-  user asked to be taken somewhere — never uninvited mid-task. The agent
-  is a hand pointing, not a hand on the wheel; there is never a state the
-  user has to escape from. (Same rule already governs approval cards —
-  inline and non-blocking, §6.5.)
-
-Mechanism — three tools plus one client loop, no screenshots, no vision
-model in the loop:
-
-1. **See**: on route change the web app pushes a structured `uiSnapshot`
-   (visible interactive elements + labeled regions as
-   `{ref, label, role, selector}`, ~120-element cap, agent chrome
-   excluded) as a `context` WS frame; a `get_current_app_context` tool
-   returns it. The screenshot→upload→"what am I looking at" round-trip
-   disappears because the app narrates itself, locally.
-2. **Navigate**: `navigate_app` validated against a curated route catalog
-   (`data-spotlight` anchors + route map, PortfolioManager's
-   `ui-catalog.json` pattern) → `ui_navigate` frame → the browser routes.
-   "Take me to Quantize" literally clicks the tab. Reversible, no approval
-   needed.
-3. **Point**: `spotlight_ui` locates a control by ref/label/catalog-id,
-   optionally navigates first, and highlights it with a hint popover
-   (hand-rolled backdrop-cutout overlay in our tokens, ~80 lines —
-   no vendored tour library needed), auto-dismissing.
-
-Why it's ours to win: (a) our web chat IS a pi agent over WS — the port is
-mechanical (`pi-web.ts` tools + two frames + a snapshot builder in the
-app); (b) it resolves the Developer-toggle discoverability tension §8
-creates — hidden capability becomes conversationally discoverable ("can
-this make a smaller model?" → reveal, navigate, spotlight); (c) the
-escalation nobody else can ship: approval-gated `ui_act` tools (fill this
-form, start this job) rendered through the §6.5 approval-card trust
-surface — guided *actions* with visible allow/deny, where PortfolioManager
-deliberately stops at navigate+point and generic computer-use agents act
-with no in-app trust model at all; (d) with the memory vault, "take me to
-that adapter I trained last week" actually resolves.
-
-Scope guard: the snapshot never includes cross-origin content; the three
-tools live on the pi-web bridge only (never the OpenAI/Anthropic API
-surface); `ui_act` is Phase 5 and requires the Phase-2 approval surface
-first.
+## History
+
+- 2026-06-12 — minimal `/chat` page (SSE against `/v1/chat/completions`).
+- 2026-06-14 — chat replaced with the Pi embed over `/ws/chat`
+  (`src/pi-web.ts`), unified hash-routed SPA with the Developer tabs.
+- 2026-06-16 — adapter discovery endpoint, web selector, and the Pi CLI
+  extension (`docs/design/web-chat-redesign.md` written; its web half now lives in
+  §3.6, the CLI half belongs to the CLI reference). The Tauri desktop-shell
+  idea (`docs/design/web-chat-redesign.md`: separate repo, sidecar `mlx-bun serve`,
+  WKWebView on `/#/chat`) was written in this period; it is tracked in
+  neither PLAN.md nor STATUS.md and is not a plan — the only desktop-specific
+  item kept here is that a shell must wrap the same page, no bespoke UI.
+- 2026-07-06 — this plan merged three design visions and the market survey;
+  Phase 0 (UI-only bug pass) and Phase 1 (block-memoized render,
+  highlighting, full sampler popover, perf strip + lane, message actions,
+  drawer, theme, shortcuts, a11y) landed. Beat matrix written (Appendix A).
+- 2026-07-07 — Phase 2 (typed module split, memory panel + REST, adapter
+  routing table, system prompt + presets, approval gate wired, Developer
+  IA) and Phase 3 (files RAG, Model Hub, Canvas, per-turn sampling scope,
+  self-healing tool calls + loop hygiene, app-aware assistant v1, command
+  palette, full-text search, export, PWA) landed.
+- 2026-08-23 — consolidated into this canonical doc; verified against src.
 
 ---
 
-## 7. Tech-stack & architecture decision
-
-**Recommendation: stay vanilla-single-file, but stop hand-rolling
-everything.** Do not migrate to React/Svelte/a SPA framework. Reasoning:
-
-- The research is explicit that every framework-based pattern found
-  (Streamdown's block memoization, react-virtuoso's windowing,
-  Radix/Base-UI's accessible primitives) assumes a component-diffing
-  runtime — but the *ideas* (memoize completed blocks, window a long list,
-  trap focus in a popover) port cleanly into vanilla JS without importing
-  the library. The current file already proves this pattern works (the
-  hand-rolled markdown renderer is competitive).
-- A framework migration is a multi-month distraction that trades a
-  self-contained, no-build, no-CDN served single file (the current
-  `app.html` header comment's own stated constraint) for a build pipeline,
-  for a benefit (component reuse, cleaner state management) that matters
-  more at team scale than at this file's actual complexity once the
-  concrete gaps (memoized rendering, syntax highlighting, math, a11y) are
-  closed directly.
-- The three visions converged independently on the same conclusion ("no
-  React needed, just a smarter diff in the existing vanilla renderer") —
-  treat that convergence as signal, not coincidence.
-
-**What does change:**
-- **Split `app.html` into modules that build into the single served file**,
-  not a runtime framework. At 2737 lines and growing (memory panel, adapter
-  routing table, perf strip, message actions are all net-new surface), a
-  single file becomes a real maintenance cost even without a framework.
-  Introduce a minimal build step (`bun build` with inline output, still
-  zero external CDN dependency at runtime) that concatenates
-  `src/web/{shell,chat,memory,adapters,quantize,finetune,dataset,status}.{html,css,ts}`
-  into the one file the server ships. This preserves every runtime
-  constraint (self-contained, no CSP-breaking externals) while making the
-  source tree navigable.
-  **The split's primary payoff is the typed contract, not navigability**
-  (amended 2026-07-06): once the inline script is TypeScript, it imports
-  `ServerMessage`/`ClientMessage` from `src/pi-web.ts` and the job-stream /
-  API response shapes from their source modules — and the single largest
-  empirical bug class disappears at compile time. Look at what Phase 0
-  actually fixed: `e.lr` vs `e.learning_rate`, array-truthiness on a WS
-  frame, a response-shape mismatch, a dead string-enum — every one is
-  contract drift between untyped inline JS and typed TS, and every one
-  becomes a build error after the split. **Scheduled: the split is Phase 2's
-  first work item** — before the memory panel and routing table add major
-  new surface to the monolith, not after.
-- **Frontend regression harness** (amended 2026-07-06 — this repo
-  oracle-gates everything else; the UI can't be the one untested surface):
-  (a) formalize the inline-script syntax check (extract `<script>`,
-  `new Function()` it) as a test/hygiene-gate step instead of an ad-hoc
-  command; (b) a `tests/web-app.test.ts` DOM-level unit file (happy-dom or
-  equivalent) covering the pure client functions — `splitBlocks`/streaming
-  memoization (streamed-then-finalized output must equal one-shot
-  `mdToHtml` on a fixture corpus: the renderer's own parity oracle),
-  `renderQueue`, `api()` error unwrapping, esc() discipline on
-  interpolation sites; (c) post-split, type-checking the frontend modules
-  in the same `tsc --noEmit` pass as the server. Playwright-class
-  end-to-end stays out of scope until there's a model-free server stub to
-  drive it against.
-- **Inline, self-hosted Shiki (or a lighter alternative) and KaTeX** as
-  vendored, no-CDN assets bundled at build time — a real dependency-weight
-  tradeoff against the "no build, no CDN" constraint, but justified because
-  monochrome code and raw-LaTeX-backslash output are concrete, embarrassing
-  gaps (§2.2 #9) with no vanilla-hand-rolled alternative that doesn't
-  reinvent a TextMate-grammar highlighter from scratch.
-- **Hand-roll block-memoized streaming and message-list windowing** rather
-  than adopt react-virtuoso — port the *idea* (completed-block skip-reparse,
-  windowed DOM recycling anchored to bottom scroll) as vanilla functions.
-
-**Tauri desktop compatibility:** the Tauri shell today wraps "the same page,
-no bespoke UI" (`docs/design/tauri-desktop-app.md:50-51`). Keep that
-contract — every feature in this plan ships to both surfaces for free by
-construction, since nothing here depends on a browser-only API. The one
-Tauri-specific opportunity (deferred to Phase 5, not required): a global
-hotkey + floating companion window (mirroring ChatGPT macOS's Option+Space
-and mlx-serve's Ctrl-Space quick-launcher) — a concrete way for the desktop
-wrap to earn native-app status instead of being a browser tab in a frame,
-made cheap by the vision pipeline already existing server-side.
-
-**Theming:** add `prefers-color-scheme` handling and a light theme (dark-only
-today, hardcoded `--bg:#000`). Respect `prefers-reduced-motion` on the
-ambient bloom/shimmer animations.
-
-**Responsive/mobile/PWA:** mobile drawer (§5.7); add a web app manifest +
-minimal cache-first service worker for the app shell (installable
-"add to home screen," zero-cost given the app is already a single served
-page) — explicitly scoped as *shell-instant-load + installability*, not
-meaningful offline chat (the app is useless without the local inference
-server running, so deep offline-caching strategy is not worth over-investing
-in). Test standalone-mode viewport height explicitly since browser chrome
-disappearing changes available height in ways normal responsive testing
-misses.
-
-**A11y:** add `role="log" aria-live="polite" aria-atomic="false"` on the
-message thread, throttled to paragraph/sentence boundaries (not per-token —
-combining focus-move and live-region announcement in the same update is the
-documented anti-pattern per WCAG ARIA23), verified against actual VoiceOver
-testing rather than one blog post's guidance. Full keyboard tab order
-through composer/sidebar/popovers, visible focus rings, Escape-to-close on
-every popover with focus returned to trigger. Ship a `Cmd/Ctrl+/` shortcut
-sheet matching ChatGPT's vocabulary where sensible (`Cmd/Ctrl+Shift+O` new
-chat, `Shift+Esc` focus composer, `Cmd/Ctrl+Shift+C` copy last response) —
-zero cost to originality, closes an embarrassing baseline gap, and "copy
-last response" reuses the existing copy-button code.
-
-**Long-thread performance:** windowed rendering (anchored-to-bottom DOM
-recycling) once a session exceeds ~150-200 rendered turns — the
-memory-flywheel/long-session north star means power users will accumulate
-very long sessions over months, more likely here than in a typical
-stateless chatbot, so this failure mode (documented against Open WebUI as
-issue #23067) is a higher-probability event for us than for competitors.
+## Appendix A — beat matrix (condensed)
 
----
+Every capability a user judges a local chat app by, judged against the best
+current implementation in the surveyed field (mlx-lm, mlx-optiq, Ollama,
+Open WebUI, LM Studio, Odysseus, Jan, Msty, GPT4All, AnythingLLM,
+Enchanted, macMLX/mlx-serve, ChatGPT, Claude.ai). **MATCH** = parity is
+enough; **BEAT** = structurally better because of something only we have;
+**SKIP** = the engine lacks the substrate, stated so silence is never an
+oversight. State column verified against src 2026-08-23. (Caveat: the
+Odysseus rows rest on a public-web identification of that product; no
+verdict flips on them.)
 
-## 8. Information architecture
-
-**Chat is the product. Everything else is a developer tool.** Collapse the
-current 7 hash-routed tabs (Chat / Quantize / Fine-tune / Build Dataset /
-Status / Curves / Routes) behind a single top-level split:
-
-- **Chat** (default, always visible) — the entire app for a first-time or
-  casual user. Contains, reachable via sidebar/settings without leaving the
-  view: Memory panel, adapter routing, sampling/structured-output advanced
-  settings, model switcher.
-- **Developer** (one nav toggle, LM Studio's User/Power-User/Developer
-  pattern collapsed to a binary since we don't need three tiers yet) —
-  reveals Quantize, Fine-tune, Build Dataset, Status, Curves, Routes as a
-  sub-nav. Persisted as a local preference (don't re-ask every session).
-
-This is not hiding capability — it's matching disclosure to audience. A
-user who never intends to quantize a model should never see that tab; a
-developer flips one switch once and gets everything, including the existing
-tabs completely unchanged in their internal design. The app-aware assistant
-(§6.6) is the escape hatch that makes the collapse safe: hidden capability
-stays conversationally discoverable — "can this make a smaller model?" and
-the agent reveals the Developer tabs, navigates, and spotlights the control,
-so progressive disclosure never becomes buried capability (risk #4's
-strongest mitigation). Status's RAM/KV "Memory"
-card keeps its current meaning (system memory) — the personal-memory vault
-gets its own distinctly-named panel (§5.5) precisely to avoid the homonym
-confusion the audit flagged.
-
-Routes (currently an iframe of a DAG, 404s in the compiled binary per
-`web-ui-pass-plan.md` #17) and Curves (a separate full-page route, never
-highlights in nav per #19) both move under Developer and get their standing
-bugs fixed as part of the IA move (§9 Phase 1).
+**Axis 1 — Chat fundamentals**
 
----
+| Capability | Verdict | State |
+|---|---|---|
+| Streaming with a prominent Stop | MATCH | shipped |
+| Stop aborts in-flight tool work, not just tokens | MATCH | shipped (verified via Pi abort) |
+| Disk persistence, no account | MATCH | shipped (Pi sessions) |
+| Regenerate | MATCH | shipped |
+| Edit-and-resend as a sibling branch, `< i/n >` | MATCH | shipped |
+| Full conversation tree view | BEAT | open |
+| Live steering + follow-up queue while streaming | BEAT | shipped |
+| Keyboard reach for stop / regenerate / new chat | MATCH | shipped |
+| Full-message copy | MATCH | shipped |
+| Delete a single message | MATCH | open |
+| Compaction seam marked in the thread | MATCH | open |
+
+**Axis 2 — Rendering**
+
+| Capability | Verdict | State |
+|---|---|---|
+| GFM markdown (hand-rolled, competitive) | MATCH | shipped |
+| Syntax highlighting, vendored | MATCH | shipped |
+| LaTeX / KaTeX | MATCH | open |
+| Mermaid | MATCH (low priority) | open |
+| Block-memoized streaming (no O(n²) re-parse) | BEAT | shipped |
+| Sandboxed HTML / SVG preview with view-source | MATCH | shipped (Canvas v1) |
+| Persistent Artifacts side panel | MATCH (opt-in) | open |
+| logprobs confidence overlay | BEAT | open (server side exists) |
+| Structured-output / grammar picker | BEAT | open (server side exists) |
+
+**Axis 3 — Model discovery and management**
+
+| Capability | Verdict | State |
+|---|---|---|
+| Hardware-fit verdict per model, measured for this Mac | BEAT | shipped (picker + Hub) |
+| Browse / search / download / load in one panel | BEAT | shipped, except live load (restart command) |
+| Capability badges (vision, quant tier) | MATCH | shipped (Hub chips) |
+| In-chat model switch without restart | MATCH | open (no in-process swap seam) |
+| Swap preserves session history | BEAT | open (same blocker) |
+| Resident multi-model pool | MATCH | open |
+| Running-models resource table with adapter / KV tier / draft columns | BEAT | open |
+| Per-layer quantization sensitivity visualization | BEAT | open |
+| `mlx-bun launch <tool>` one-command coding-tool wiring | BEAT | open (CLI) |
+
+**Axis 4 — Prompts, presets, sampling**
+
+| Capability | Verdict | State |
+|---|---|---|
+| Full sampler set in the UI | MATCH | shipped (`logit_bias` not exposed) |
+| Per-model correct defaults | MATCH | shipped (`ready.genDefaults`) |
+| Per-message sampling scope | MATCH | shipped (`next_turn`) |
+| Saved presets (prompt + sampling) | MATCH | shipped v1 (local only) |
+| Slash-command prompt library with variables | MATCH | open |
+| Model-string variant shorthand (`:no-think`) | MATCH | open |
+| Persona bundles bound to a real adapter swap | BEAT | open |
+| Explicit "warm this session" | MATCH | open |
+
+**Axis 5 — Documents and RAG**
+
+| Capability | Verdict | State |
+|---|---|---|
+| Zero-config attach-and-answer | MATCH | shipped |
+| Transparent inline-vs-retrieval decision | MATCH | shipped (threshold in `rag.ts`) |
+| `[n]` citations + Sources panel | MATCH | shipped |
+| Unified `#` mention over files + vault | MATCH | shipped |
+| Per-document retrieval-mode toggle | MATCH | open |
+| Hybrid BM25 + vector + rerank | SKIP (for now) | open, deferred tier |
+| Semantic search over the user's own chat history | BEAT | partial (full-text body search shipped; semantic open) |
 
-## 9. Phased roadmap
-
-Each phase is independently shippable and includes the relevant tactical
-fixes from `docs/archive/planning/web-ui-pass-plan.md` by number. Ordered by
-effort/reward, starting with free wins, ending with differentiator
-features. The [beat matrix](web-chat-beat-matrix.md) is the coverage
-checklist across all phases: at any release point, every matrix row is
-either shipped, in a scheduled phase below, or a written SKIP — a MATCH row
-with none of those three states is a planning bug.
-
-### Phase 0 — Bug fixes, one UI-only PR (no server changes, ~2-3 days)
-
-Ship first, blocks nothing, touches no backend:
-
-- `web-ui-pass-plan.md` #1: fine-tune live stats field-name mismatch
-  (`e.lr`/`e.tps` → `e.learning_rate`/`e.tokens_per_sec`).
-- #2: queue-bar array-truthiness bug (`.length` guards, `app.html:1683-1684`).
-- #4: `[object Object]` adapter-mount error (unwrap `data.error.message` in
-  `api()`, `app.html:1080`).
-- #5: quantize inspect result parsing (`support: boolean`, not string enum).
-- #6: stale "switch to 12B for vision" copy → model-agnostic.
-- #15 (partial): escape `a.id`/`a.path` interpolation in the adapter
-  dropdown (`app.html:1897-1898`) — this is a live unescaped-HTML injection
-  point, fix immediately regardless of the larger adapter-UI rework in
-  Phase 2.
-- #16: delete dead `src/status-page.html` + `scripts/status-page-stub.ts`.
-- #18: CLI banner copy fix (`src/cli.ts:64`).
-
-### Phase 1 — Free wins: wire existing backend, fix the render bottleneck (1-2 weeks)
-
-The highest-leverage phase — almost all wiring of existing capability +
-one architectural render fix. Two deliberate, thin server additions ride
-along (scoped in §2.3's caveat): the per-turn `lane` field for the perf
-strip, and the gc plan/execute endpoints (#9):
-
-- **Streaming render fix** (§5.3): block-memoized append, replacing
-  `appendDelta`'s full re-parse. This unblocks every other visual change in
-  this phase since they all touch the same render path.
-- Syntax highlighting (vendored Shiki or equivalent) + KaTeX, applied to
-  completed blocks only.
-- Sampling popover: full parameter set (`min_p`, XTC, penalty families +
-  context windows, `logit_bias`, seed) behind an Advanced disclosure;
-  per-model `generation_config.json`-sourced defaults (fixes
-  `web-ui-pass-plan.md` #14 and #8). Wire end-to-end: popover → `set_sampling`
-  WS frame → `SamplingOverrides` → `injectSampling` (all three currently
-  hardcode 3 fields, `pi-web.ts:171-176`).
-- Perf strip: tok/s, TTFT, context-fill, lane badge (serial/spec-decode/batched)
-  under the composer. Server side: add the per-turn `lane` field to the
-  usage payload + pi-web frame (see §2.3 caveat) — the badge must be
-  server-driven, never client-inferred (risk #5).
-- Message actions: copy full assistant message, regenerate, edit-message-as-branch
-  with `< 1/2 >` toggle (reuses `SessionManager.forkFrom`).
-- Adapter dropdown → chip with specialty description + `compatible` graying
-  (still single-select in this phase; full routing table is Phase 2).
-- `web-ui-pass-plan.md` #3 (library staleness), #7 (`sft_scope` control),
-  #9 (gc affordance — needs the new `/api/gc/plan`/`/api/gc/execute`
-  endpoints), #11-13 (the queued `listCanonical`/`visionCapable`/`supportTier`
-  wiring, land as its own small PR once server-side lands), #20 (library
-  "too big" row fix), #21 documented as a known limitation.
-- Mobile sidebar → slide-over drawer.
-- Session search box.
-- `prefers-color-scheme` + light theme + `prefers-reduced-motion` guards.
-- `Cmd/Ctrl+/` shortcut sheet + ChatGPT-vocabulary bindings.
-- `role="log" aria-live="polite"` on the thread, focus-trap/Escape on
-  popovers.
-
-### Phase 2 — Module split, memory surface + adapter routing table (2-3 weeks)
-
-The two headline differentiators, now that rendering is fixed and settings
-have room — preceded by the structural fix that keeps the growing surface
-honest:
-
-- **Module split FIRST** (§7, amended): split `app.html` into
-  build-concatenated `src/web/*.ts` modules (one served file unchanged at
-  runtime) so the frontend imports `ServerMessage`/`ClientMessage` from
-  `src/pi-web.ts` and job/API shapes from their source modules — the
-  Phase-0 bug class (contract drift) becomes a compile error before the
-  memory panel and routing table add major new surface. Ship the frontend
-  regression harness (§7) in the same pass: formalized script-syntax gate,
-  `tests/web-app.test.ts` over the pure client functions (streaming
-  memoization vs. one-shot `mdToHtml` parity fixtures, `renderQueue`,
-  `api()` unwrapping), frontend modules in the same `tsc --noEmit` pass.
-- **Per-chat system prompt** (beat matrix Axis 4 — table stakes in every
-  competitor, currently impossible: the web chat's prompt is hardcoded in
-  `buildPiAgentSurface`/`buildWebChatSystemPrompt`): an editable
-  system-prompt field per session, layered over the built-in surface prompt,
-  carried on the WS session; **presets v1** (named system-prompt + sampling
-  bundle, saved locally) shaped so Phase 5's persona bundles (adapter +
-  tools + memory scope) are a thin extension.
-- **Session auto-titling** (Axis 11): v1 derives from the first user
-  message; model-generated titles optional and never queued ahead of real
-  turns.
-- **Vault-init endpoint** (§5.1 prerequisite): the first-run consent card
-  creates the vault from the web UI — needs a thin
-  `POST /api/memory/init` wrapping what `mlx-bun memory init` does in the
-  CLI (today it's CLI-only and invisible to non-terminal users).
-- Memory REST wrappers (prerequisite, server-side — see §2.3 caveat):
-  `GET /api/memory/status|list|search|article` + a git log/diff read,
-  thin delegating wrappers over `src/memory/vault.ts` / `tools.ts`
-  internals. The agent-tool surface stays read-only and unchanged.
-- Memory panel (§5.5): vault browser, search, per-article git-diff view,
-  status strip, `Reference/` sub-view, wikilink graph via `memory_links`.
-- Memory provenance chips on assistant turns that consulted memory tools
-  (distinct card style from generic tool calls, §5.4).
-- Personalized hero/empty-state chips driven by `vaultStatus()` +
-  `/v1/adapters/available` (§5.1); inline memory-consent card for first run.
-- Adapter routing table: three-state visibility (available/loaded/selected),
-  RAM cost per loaded adapter, `"a+b"` stacking support in the picker.
-- Model switcher: clickable `#nav-model`, hardware-fit badge via `GET /fit`,
-  session-preserving swap.
-- Tool-approval gate properly wired: `codingTools: true` opt-in toggle in
-  settings, editable arguments before approve/deny, durable tool-identity-keyed
-  "always allow" persistence.
-- Structured-output picker (JSON-schema textarea + live validation) in the
-  Advanced disclosure.
-- Unified `#` retrieval mention (files + vault as one picker).
-- Information architecture move: collapse Quantize/Fine-tune/Build
-  Dataset/Status/Curves/Routes behind the Developer toggle (§8); fix
-  `web-ui-pass-plan.md` #17 (Routes 404 in compiled binary) and #19 (Curves
-  nav highlight) as part of this move.
-
-### Phase 3 — Concede nothing: the table-stakes sweep (2-3 weeks)
-
-(Added 2026-07-06 under the superset doctrine — the beat-matrix MATCH rows
-a daily LM Studio / Open WebUI / optiq-Lab user would immediately miss.
-Each item names its matrix axis.)
-
-- **Chat-with-files RAG v1** (Axis 5): attach-and-answer over attached
-  documents with `[n]` citation markers + a sources panel — optiq Lab's
-  BM25 dependency-free bar is the v1 target; we have `src/embed.ts` for a
-  vector upgrade later, but don't block v1 on it. Extend the unified `#`
-  mention (Phase 2) so files and vault articles are one retrieval gesture.
-- **Model Hub** (Axis 3): in-app browse (published quants / HF search /
-  local) → download with progress → one-click load-to-server + new-chat,
-  session-preserving hot-swap — optiq's Hub flow, PLUS the `/fit`
-  hardware-verdict column per row that optiq lacks. "Install to chat under
-  a minute" stops dead-ending at the CLI.
-- **Canvas v1** (Axis 2): fenced-HTML blocks rendered in a sandboxed
-  iframe with a view-source toggle (optiq's shipped pattern; the full
-  Artifacts side-panel stays Phase 5).
-- **Per-message sampling scope** (Axis 4): apply-to-next-turn-only option
-  in the sampling popover (optiq's per-message temp/max-tokens/thinking
-  granularity), on top of the session-level overrides from Phase 1.
-- **Self-healing tool-call loop** (Axis 7): investigate what the pi SDK
-  already retries; add format-aware malformed-call repair (we own the
-  per-model parsers in `src/tool-call.ts`), consecutive-duplicate dedup, a
-  retry budget, and a loop cap with force-finish — small local models
-  malform calls more than cloud frontiers, so this matters more here than
-  it does for optiq.
-- **Stop kills tool subprocesses** (Axis 1): verify/wire abort through the
-  pi tool layer so Stop ends running bash/web-fetch work, not just the
-  token stream.
-- **Chat export** (Axis 10): Markdown/plain-text per session; portable
-  JSON export/import (sessions are already local JSON on disk).
-- **Command palette** (Axis 10, `Cmd/Ctrl+K`): new chat / session search /
-  model + adapter switch / thinking toggle / theme.
-- **Full-text session search** (Axis 10 BEAT): body search, not
-  title-only — no cloud indexing-cost excuse applies locally.
-- **PWA installability** (Axis 10): manifest + shell service worker
-  (installability + instant shell, explicitly not offline chat).
-- **App-aware assistant v1** (§6.6, matrix Axis 12 — promoted from Phase 4
-  2026-07-06: best wow-per-effort in this phase): `uiSnapshot` context push
-  on route change + a compact ambient context line (route/view/step)
-  auto-attached to every turn, `get_current_app_context` + `navigate_app` +
-  `spotlight_ui` tools on the pi-web bridge, curated route/anchor catalog,
-  hand-rolled spotlight overlay choreographed with the streaming
-  explanation. Pattern proven in `~/Code/PortfolioManager` on the same pi
-  SDK. The ambient every-view floating panel and process-state snapshot
-  enrichment mature in Phase 4; approval-gated `ui_act` stays Phase 5.
-
-### Phase 4 — Trust & speed differentiators (1-2 weeks)
+**Axis 6 — Memory and personalization**
 
-- "Temporary chat" pill — provably never written to disk.
-- Per-session disk-touch tally (files read/written this session) as a
-  trust log distinct from tool-call cards.
-- Prompt-cache-hit telemetry surfaced per turn ("served from cache").
-- logprobs-derived confidence visualization (optional, behind Advanced) —
-  wire the existing `logprobs`/`top_logprobs` server support into an
-  opt-in per-turn confidence overlay.
-- HLG tone-curve controls folded into the main chat composer (currently
-  disconnected, only reachable via the standalone Curve Designer page).
-- **App-aware assistant: ambient panel maturity** (§6.6): the floating
-  assistant affordance on every view (Developer tabs included) with the
-  Chat tab as the deep surface, and process-state enrichment of the
-  snapshot (wizard step, job progress, field completion) — building on
-  Phase 3's v1.
-
-### Phase 5 — Stretch differentiators (ongoing, pick 1-2 per quarter after Phase 4)
-
-The next tier of unmistakable bets:
-
-- Full conversation tree view (not just linear `< 1/2 >` toggle) — a
-  genuine gap neither ChatGPT nor Claude has shipped natively.
-- Persona/bundle system: named (system prompt + adapter + tool allowlist +
-  memory scope) bundles, switchable in one click, exportable/importable.
-- Compare/split view: same prompt against two adapters or two quant tiers
-  simultaneously (Msty's Split Chat pattern, repointed at LoRA A/B testing
-  instead of cross-vendor model shopping).
-- Overnight/background training loop with a persistent monitor and
-  completion notification ("wake up to a new adapter") — closes
-  `docs/design/adapters-end-to-end.md` §D, currently PENDING.
-- Tauri-specific: global hotkey + floating companion window with instant
-  screenshot capture (vision pipeline already exists).
-- Opt-in-by-heuristic Artifacts/Canvas side-panel for long code/HTML blocks
-  with highlight-to-scope-edit (Canvas v1 iframe ships in Phase 3).
-- MCP client support (beat matrix Axis 7 — table stakes for 2026 agentic
-  chat per Open WebUI/LM Studio/Jan; moderate-large build behind the
-  existing tool plumbing, scheduled here rather than silently absent).
-- Skills-style self-evolving procedural memory (Odysseus's pattern; large
-  separate system, matrix SKIP-for-v1 — revisit once the memory panel has
-  usage data).
-- Mermaid diagram rendering (matrix Axis 2 — add if the code-heavy audience
-  demands it; completed-block hook makes it cheap to slot in).
-- Vector upgrade to chat-with-files RAG (BM25 v1 → `src/embed.ts` hybrid).
-- **Approval-gated `ui_act` tools** (§6.6 escalation): the app-aware
-  assistant graduates from navigate+point to fill-this-form /
-  start-this-job, every action rendered through the approval-card gate
-  with argument preview — guided actions with a visible trust surface,
-  which neither PortfolioManager (stops at navigate+point) nor generic
-  computer-use agents (act with no in-app trust model) ship. Requires
-  Phase 2's approval surface + Phase 4's v1.
+| Capability | Verdict | State |
+|---|---|---|
+| Visible, searchable memory (vault browser) | BEAT | shipped |
+| Per-turn provenance chip | BEAT | shipped |
+| Git history / diff of what synthesis changed | BEAT | shipped |
+| Wikilink graph view | BEAT | open (links line only) |
+| Scope control (session / surface / global) | MATCH | open |
+| Pause vs reset as distinct actions | MATCH | open |
+| Memory graduating into trained weights, marked in UI | BEAT | roadmap — never claim early |
+| First-run consent in the hero | BEAT | shipped |
+| Synthesis schedule status visible | MATCH | shipped (status strip) |
+| Self-evolving procedural "Skills" | SKIP (v1) | open, deferred |
+
+**Axis 7 — Tools and agents**
+
+| Capability | Verdict | State |
+|---|---|---|
+| Tool calls as distinct cards | MATCH | shipped |
+| Inline non-blocking approval cards | MATCH | shipped |
+| Editable arguments before approval | MATCH | shipped |
+| Durable tool-identity-keyed always-allow | BEAT | shipped (`src/tool-approvals.ts`) |
+| Gated tools reachable in web chat (opt-in) | MATCH | shipped |
+| Sandboxed Python tool | MATCH | open |
+| Self-healing malformed tool calls | MATCH | shipped (`src/tool-call.ts`) |
+| Loop hygiene (dedup, retry budget, cap, force-finish) | MATCH | shipped |
+| Argument / diff preview before approval | BEAT | shipped (args); file diff open |
+| Per-session disk-touch tally | BEAT | open |
+| Curated default tool pack on first run | MATCH | partial (read, web_search, memory, app-aware) |
+| MCP client | MATCH | open |
+| Web search + URL fetch without keys | MATCH | `web_search` shipped; `web_fetch` not in the web allowlist |
+| Approval-fatigue-aware defaults (read-only auto-allowed) | MATCH | shipped |
+| `tool_choice` forcing in the chat UI | SKIP | API only by design |
+| Visual agent-flow builder | SKIP | not planned |
+| Deep-research trace panel | MATCH | open |
+
+**Axis 8 — Multimodal**
+
+| Capability | Verdict | State |
+|---|---|---|
+| Image attach + inference | MATCH | shipped (gated by `ready.vision`) |
+| No text-only tax from the vision path | MATCH | verify |
+| Image generation | SKIP | no substrate |
+| Speech-to-speech voice | SKIP | no substrate |
+| Native OS dictation / TTS | SKIP | no substrate |
+| Document attach (PDF / DOCX) with inline-vs-RAG | MATCH | text files shipped; binary formats open |
+| Global hotkey + floating companion window with screenshot | BEAT | open (needs a desktop shell) |
 
----
+**Axis 9 — Performance and transparency**
 
-## 10. Success metrics & risks
-
-### Success metrics
-
-- **Memory visibility:** % of sessions where a memory-consulting turn shows
-  a provenance chip (should approach 100% of turns where memory tools
-  actually fired — a wiring correctness metric, not an engagement metric).
-- **Adapter routing usage:** adapter-swap events per week per active user,
-  post-Phase-2 vs. pre-redesign baseline (currently near-zero given the
-  dropdown's poor legibility).
-- **Render performance:** p95 time-to-paint for a 2000-token streamed
-  response, measured on the M4 Pro reference machine per the `benchmark.sh`
-  discipline — before/after the block-memoization fix, quotable only from a
-  preflight-gated run.
-- **Tool-approval engagement:** % of gated-tool calls that get an explicit
-  approve/deny vs. auto-deny timeout (a proxy for whether the trust surface
-  is legible enough to act on, not just present).
-- **Mobile session continuity:** % of mobile sessions where a user
-  successfully switches sessions via the new drawer (currently impossible).
-- **A11y baseline:** VoiceOver walkthrough of a full send→stream→tool-call→
-  approve cycle completes with no unannounced state changes.
-
-### Top risks & mitigations
-
-1. **Scope creep into a framework rewrite.** Mitigation: §7's explicit
-   recommendation against a SPA framework migration, with the reasoning
-   documented so a future session doesn't relitigate it without new
-   evidence.
-2. **Memory panel becomes a context-dump UX that violates "quiet retrieval"
-   (Design Principle 4).** Mitigation: provenance chips are collapsed by
-   default, expand-on-click only; no chat-thread content is duplicated into
-   the panel automatically.
-3. **Adapter routing table adds setup friction before first chat**, cutting
-   against "install to chat under a minute." Mitigation: the routing table
-   is progressive disclosure behind settings; the composer chip default
-   ("general-purpose", no adapter) requires zero configuration.
-4. **The IA collapse (Developer toggle) hides a capability an existing user
-   relies on daily** (e.g. someone who lives in Quantize). Mitigation: the
-   toggle persists as a local preference and is one click, not a
-   re-onboarding flow; existing deep-links (`#/quantize` etc.) keep working
-   unchanged.
-5. **Perf-strip lane badge becomes stale or wrong** during a mode
-   transition (e.g. spec-decode falling back to serial mid-session).
-   Mitigation: badge is server-driven per-turn from the same instrumentation
-   already used for tok/s, not client-inferred; treat any mismatch as a P0
-   bug since the entire differentiator rests on it being trustworthy.
-6. **The tool-approval "always allow" persistence reintroduces the exact
-   fragility documented against Claude Desktop** (resets on update). Mitigation:
-   store approvals in a versioned, file-backed config keyed by stable tool
-   identity from day one, with an explicit migration test in CI across a
-   simulated version bump.
-7. **Memory git-diff view surfaces something the user finds unsettling**
-   (seeing exactly how much was inferred about them). Mitigation: frame the
-   diff view as user-initiated inspection, not a push notification; pair
-   with the existing pause/reset actions (already a good ChatGPT/Claude
-   pattern) so the "off switch" is never more than one click from the
-   panel that shows the most.
-8. **(Added 2026-07-06) The superset doctrine balloons scope until nothing
-   ships.** Matching 116 matrix rows is a quarter-plus of work; the
-   original wedge-only framing failed by conceding too much, and the
-   correction can fail by attempting everything at once. Mitigation: the
-   phases stay independently shippable and strictly ordered
-   (fundamentals → structure → table-stakes sweep → trust/speed →
-   stretch); within Phase 3 the sweep is ordered by "what would a daily
-   competitor user miss first"; the matrix's shipped/scheduled/SKIP states
-   are re-audited at each phase boundary rather than renegotiated
-   mid-phase.
-9. **(Added 2026-07-06) The untested, untyped frontend erodes as it
-   grows.** Phase 0's bug class (contract drift between inline JS and
-   typed TS) recurs with every new surface until the module split lands.
-   Mitigation: the split + frontend regression harness are Phase 2's
-   FIRST work item (§7), before the memory panel and routing table add
-   the next thousand lines; until then, every new WS frame or API shape
-   consumed by `app.html` gets a line in `tests/pi-web.test.ts` proving
-   the producing side's contract.
+| Capability | Verdict | State |
+|---|---|---|
+| Live tok/s + TTFT | MATCH | shipped |
+| Context-window meter | MATCH | shipped |
+| Serving-lane badge in the live chat | BEAT | shipped |
+| Prompt-cache-hit telemetry per turn | BEAT | open |
+| In-app benchmark tab | MATCH | open (harness exists: `scripts/bench-serve.ts`) |
+| Compare / split view (base vs adapter, parity vs fast path) | BEAT | open |
+| Concurrency / batch-slot indicator | MATCH | open |
+| `/fit` surfaced beyond Status | BEAT | shipped (picker, Hub) |
+| Auto-sized context to available memory | BEAT | open (UI); admission clamps server-side |
+| Provable zero-cloud-calls indicator | BEAT | open |
+
+**Axis 10 — Polish**
+
+| Capability | Verdict | State |
+|---|---|---|
+| Dark + light, `prefers-color-scheme` | MATCH | shipped |
+| `prefers-reduced-motion` | MATCH | shipped |
+| Shortcut sheet (Cmd/Ctrl+/) | MATCH | shipped |
+| Command palette (Cmd/Ctrl+K) | MATCH | shipped |
+| Responsive to 375 px with a drawer | MATCH | shipped |
+| PWA installability | MATCH | shipped |
+| `aria-live` thread, focus traps, Escape | MATCH | shipped |
+| Developer progressive disclosure | MATCH | shipped |
+| Server-driven personalized onboarding chips | BEAT | shipped |
+| Chat export (Markdown / JSON) | MATCH | shipped |
+| Shareable links | SKIP | scope decision (single-user, local) |
+| Full-text search across message bodies | BEAT | shipped |
+| Session pins / folders | MATCH | open |
+| Clean quit / no zombie process in a desktop shell | MATCH | n/a until a shell exists |
+
+**Axis 11 — Sessions**
+
+| Capability | Verdict | State |
+|---|---|---|
+| New / list / open / fork / delete | MATCH | shipped |
+| Session search box | MATCH | shipped |
+| Per-message branch-from-here | MATCH | shipped (edit / regenerate siblings) |
+| Projects (scoped knowledge + memory + adapter) | BEAT | open (persona bundles) |
+| Temporary / incognito chat, provably not on disk | BEAT | open |
+| Multi-adapter compare | BEAT | open |
+| Portable JSON export / import | MATCH | export shipped; import open |
+
+**Axis 12 — App-aware assistant**
+
+| Capability | Verdict | State |
+|---|---|---|
+| Agent sees the live UI (structured snapshot) | BEAT | shipped |
+| Agent navigates the app | BEAT | shipped |
+| Agent spotlights a control | BEAT | shipped |
+| Approval-gated agent UI actions (`ui_act`) | BEAT (stretch) | open |
+
+**The beats** (ranked by visible wow × structural defensibility): lane
+badge in live chat · memory git-diff view · per-turn provenance chip ·
+legible, stackable adapter routing · base-vs-adapter compare view · model
+swap that preserves the session · prompt-cache-hit telemetry · provably
+local temporary chat · diff preview before approval · full-text body
+search · the app-aware assistant. The first four and the last two are
+shipped; the rest are §5.