feat(quick-pane): add follow-up chat renewal

docs/quick-pane-v1-design.md

@@ -2,13 +2,13 @@
 
 ## Goal
 
-Quick Pane is a small native window opened by a desktop/window-manager hotkey. It accepts one short request and returns either a useful answer or an explicit escalation within 60 seconds.
+Quick Pane is a small native window opened by a desktop/window-manager hotkey. It keeps a lightweight chat for short requests and returns useful answers and offers an explicit manual full-session continuation when needed.
 
 ## Scope
 
 - Cold-launchable, independent Tauri window.
 - Compact request input, quick-model picker, and workspace picker.
-- Configured default workspace, quick model, follow-up model, supplemental prompt, time limit, and read-only tool limit.
+- Configured default workspace, quick model, follow-up model, supplemental prompt, and read-only tool-call guidance.
 - Ephemeral Quick Pane UI state and best-effort runtime cleanup.
 - Advisory read-only/no-subagent instructions, inline permission/clarification prompts, and explicit normal-session handoff.
 
@@ -25,23 +25,29 @@ Quick Pane is a small native window opened by a desktop/window-manager hotkey. I
 
 Desktop/WM configuration owns the hotkey so it works while Pi GUI is absent. Settings displays and copies the required command; it does not register or modify system bindings.
 
-The app creates or focuses a dedicated small Quick Pane window. `Esc` closes it and discards quick-only state.
+The app creates or focuses a dedicated Quick Pane window. `Esc` and **Close** hide the window without clearing an idle runtime, draft, model catalog, or transcript. Reopening within three minutes preserves that chat. Focusing an idle chat after more than three minutes closes it and prepares a fresh runtime; active work is never renewed on this timer.
 
 ## Request lifecycle
 
 1. Pane chooses configured default workspace and quick model, with per-request overrides.
 2. It starts an isolated Pi RPC runtime in the chosen workspace.
-3. Immutable prompt text instructs one request, read-only inspection, zero subagents, configured wall-clock budget, and configured tool-call budget. Current Pi RPC cannot trusted-enforce those instructions; event-derived caps and cleanup are best effort.
-4. Supplemental user instructions append beneath those prompt guardrails.
-5. Pane streams answer, tool status, and inline extension requests.
-6. On success it displays final answer. On a cap it preserves partial answer, explains the cap, and offers handoff.
+3. Immutable prompt text instructs short requests, read-only inspection, zero subagents, and configured tool-call budget. Current Pi RPC cannot trusted-enforce those instructions.
+4. Supplemental user instructions append beneath those prompt guardrails. Pane maps each exact submitted raw prompt back to entered user text, so transcript rendering never parses delimiter-like text from instructions or requests. Unknown reflected messages with the Quick Pane advisory prefix are hidden.
+5. Pane streams the full ordered user/assistant transcript, tool status, and inline extension requests. Follow-ups reuse the same runtime until explicit or stale renewal. Starting each follow-up clears only prior turn's handoff answer state; transcript messages remain visible.
+6. After first completed assistant answer, it displays a prominent manual **Continue in full session** action. Escalation and runtime-error notices remain visible and also offer handoff. No answer triggers a full session automatically.
+
+## Chat reset and layout
+
+`Ctrl+N` starts a new chat: it aborts active work when needed, closes the old runtime with retryable cleanup, clears draft/transcript and submitted-prompt mappings, then prepares a fresh runtime and model catalog. Automatic stale renewal performs the same idle-chat replacement while preserving the current draft.
+
+Quick Pane opens at 640×360 with a 520×360 minimum. Frontend content measurement requests native growth up to 520px tall after transcript updates; beyond that cap, transcript and model results scroll inside pane while outer document and pane remain clipped.
 
 ## Escalation
 
-Mutations, subagents, multi-step work, and likely-over-budget work preflight-escalate. The quick agent explains why. The user clicks **Continue in full session** to open a new normal session with:
+Mutations, subagents, multi-step work, and complex work preflight-escalate. The quick agent explains why. The user clicks **Continue in full session** to open a new normal session with:
 
 - original request;
-- escalation note or partial findings;
+- transcript-derived current-chat context and escalation note or partial findings;
 - selected workspace; and
 - configured follow-up model.
 
@@ -51,16 +57,16 @@ No full session opens automatically.
 
 Quick Pane surfaces Pi extension `select`, `confirm`, `input`, and `editor` requests inline. User explicitly answers every prompt. Existing policy remains authoritative: ordinary workspace reads may be allowed, external-directory reads may ask, and sensitive paths may deny.
 
-Quick Pane UI request, output, and transcript stay in memory only. Closing pane removes them and triggers best-effort quick-runtime cleanup. Pi may write temporary session data while the child runs. A deliberate handoff persists only in newly created normal-session history.
+Quick Pane UI request, output, and transcript stay in memory only. Hiding the pane preserves an idle chat for up to three minutes. `Ctrl+N`, stale idle renewal, app teardown, or another explicit chat replacement clears in-memory chat state and requests best-effort quick-runtime cleanup. Pi may write temporary session data while the child runs. A deliberate handoff persists only in newly created normal-session history.
 
 ## Failure handling
 
-Unavailable models, startup failures, and crashes render inline actionable errors: retry, pick another quick model, or open normal session. No silent fallback model is used.
+Settings-startup and model-catalog preparation errors render an inline **Retry** action. Retry completes failed-runtime cleanup before preparing a replacement; model selection and Send remain disabled until a valid runtime catalog is ready. Request/runtime failures offer **Continue** only when a valid request and handoff configuration exist. An empty catalog remains visible with model selection and Send disabled. No silent fallback model is used.
 
 ## Settings validation
 
-Settings reject invalid model, workspace, custom prompt, or numeric limit. Failed save keeps previously stored configuration and marks invalid fields.
+Settings reject invalid model, workspace, custom prompt, or tool-call limit. Failed save keeps previously stored configuration and marks invalid fields.
 
 ## Acceptance
 
-From cold or hot desktop state, the hotkey produces an answer or explicit escalation within 60 seconds. Automated component, bridge, persistence, and native tests cover behavior; a manual cold-launch/hotkey smoke test verifies desktop integration.
+From cold or hot desktop state, Quick Pane starts an unlimited chat without a time-based abort. Automated component, bridge, persistence, and native tests cover behavior; a manual cold-launch/hotkey smoke test verifies desktop integration.

docs/quick-pane-v1-implementation-plan.md

@@ -14,28 +14,32 @@
 
 ## 3. Build Quick Pane React surface
 
-- Add isolated state reducer/component for request, selected model/workspace, progress, final answer, partial answer, escalation, extension UI, and errors.
-- Keep pane transcript in component memory only and request best-effort runtime/session cleanup on close.
+- Add isolated state/component logic for request, selected model/workspace, progress, full ordered transcript, current-turn partial answer, escalation, extension UI, and errors. Clear prior answer/handoff state before every follow-up while retaining transcript messages.
+- Keep pane transcript and exact raw-prompt-to-visible-request mappings in component memory only. Hide unknown reflected Quick Pane advisory prompts rather than delimiter-parsing them.
+- Make `Esc` and **Close** native hide actions. Preserve idle runtime, draft, catalog, and transcript when reopened within three minutes; renew stale idle chats on focus.
+- Add `Ctrl+N` reset semantics: abort active work, close with cleanup retry, clear draft/transcript/prompt mappings, and prepare a fresh catalog/runtime.
+- Open native window at 640×360 with a 520×360 minimum; use ResizeObserver/RAF content measurement to request native growth only up to 520px tall, then use internal scrolling.
+- Add visible Retry recovery for settings-startup and model-catalog preparation failures. Finish failed-runtime cleanup before replacement preparation, and keep model/Send disabled until ready.
 - Add Settings controls and copyable WM launch command.
 
 ## 4. Add constrained bridge runtime mode
 
 - Extend runtime creation with Quick Pane metadata and ephemeral session lifecycle.
 - Build immutable advisory quick-agent instructions around supplemental custom prompt.
-- Use observed tool events and wall-clock cancellation for best-effort caps; document that current Pi RPC cannot trusted-enforce read-only/no-subagent policy or memory-only runtime sessions.
-- Instruct preflight escalation for mutation, subagent, multi-step, or likely-over-budget requests.
+- Use advisory tool-call guidance only; document that current Pi RPC cannot trusted-enforce read-only/no-subagent policy or memory-only runtime sessions.
+- Instruct preflight escalation for mutation, subagent, multi-step, or complex requests.
 
 ## 5. Route interaction and handoff
 
 - Reuse existing extension request/response protocol for inline Quick Pane permission and clarification dialogs.
-- On explicit continuation, create normal runtime using workspace and follow-up model.
-- Prefill it with original request plus escalation note/partial findings.
+- After first completed assistant answer, show manual continuation with valid handoff configuration; never open a session automatically.
+- Prefill normal session with transcript-derived current-chat context, workspace, and follow-up model.
 
 ## 6. Test and document
 
 - Add unit tests for settings validation/persistence and guardrail classification.
-- Add component tests for pane states, picker behavior, inline prompts, error/retry, and cleanup.
-- Add Rust/bridge tests for window command routing, budget cancellation, ephemeral cleanup, and handoff payload.
+- Add component tests for pane states, full transcript and exact prompt reflection, picker behavior, inline prompts, hide/renew/reset behavior, runtime and extension errors, stale-created cleanup retry, and handoff.
+- Add Rust/bridge tests for window command routing, bounded content resizing, ephemeral cleanup, and handoff payload.
 - Add manual WM hotkey/cold-launch instructions and smoke checklist.
 
 ## Sequencing

docs/quick-pane-v1-test-plan.md

@@ -6,36 +6,43 @@
 
 - Valid defaults load when no Quick Pane configuration exists.
 - Valid configuration round-trips through native storage.
-- Invalid workspace, model, prompt, time, and tool values show field errors and retain prior saved state.
+- Invalid workspace, model, prompt, and tool-call values show field errors and retain prior saved state.
 - Quick model picker affects quick request only; follow-up model remains configured default.
 
 ### Runtime guardrails
 
 - Quick runtime receives immutable advisory read-only/no-subagent prompt text plus supplemental prompt.
-- Mutation, subagent, multi-step, and likely-over-budget requests are instructed to preflight-escalate.
-- Runtime applies best-effort configured 60-second/6-tool caps from observed events and exposes reason.
-- Partial result survives a cap; pane transcript is removed and runtime cleanup requested after pane close.
+- Mutation, subagent, multi-step, and complex requests are instructed to preflight-escalate.
+- Quick Pane has no time-based abort; configured tool-call guidance remains prompt-only.
+- Full ordered user/assistant transcript remains visible for follow-ups. Each follow-up clears prior answer/partial-handoff state before model, thinking, or prompt submission.
+- Reflected user messages are resolved by exact submitted raw-prompt mapping. Supplemental instructions and entered requests containing `\n\nUser request:` render only the exact entered multiline request; unknown Quick Pane advisory prompts are hidden and ordinary user text remains visible.
 - Tests/documentation state current limitations: Pi RPC cannot trusted-enforce prompt policy or guarantee memory-only temporary sessions.
 
 ### UI and bridge
 
 - Quick Pane opens/focuses through dedicated native command without changing normal `--new` or `--toggle` behavior.
 - Workspace and quick-model picker values reach runtime creation.
-- Permission and clarification extension requests render inline and send correct response payloads.
-- Retry, model change, and normal-session actions appear for startup/model/crash errors.
-- Handoff creates normal session with original request, escalation note, workspace, and follow-up model.
+- Permission and clarification extension requests remain visible inline and send correct response payloads; runtime errors do not erase partial transcript or pending extensions.
+- `Esc` and **Close** hide without cleanup. Reopening an idle chat at or under three minutes preserves runtime, draft/catalog, and transcript; reopening after more than three minutes renews it.
+- `Ctrl+N` prevents browser default, aborts active work when needed, retries queued cleanup, clears draft/transcript/raw-prompt mappings, and prepares a fresh runtime/catalog. Idle reset does not send an abort.
+- Stale runtimes created by superseded preparation are queued after cleanup failure and retried before later stale-chat renewal.
+- Quick Pane opens at 640×360 with a 520×360 native minimum. Content-driven native growth stops at 520px tall; transcript and model results scroll internally after that.
+- Settings-startup and model-catalog errors show **Retry**. Catalog retry cleans up failed runtime before replacement preparation; model selection and Send stay disabled until ready.
+- **Continue in full session** appears after first completed assistant answer with valid handoff settings, plus preserved request/runtime escalation notices. It never starts automatically. Handoff creates normal session with transcript-derived current-chat context, workspace, and follow-up model. Current answer extraction starts only after exact mapped submitted prompt in snapshot transcript; failed follow-ups with no current assistant message never reuse prior answer as partial findings.
 
 ## Manual smoke checklist
 
-1. Configure default workspace, models, prompt, and 60-second/6-tool limits.
+1. Configure default workspace, models, prompt, and tool-call guidance.
 2. Add desktop/WM hotkey using Settings launch command.
 3. From no running GUI process, invoke hotkey and verify Quick Pane opens/focuses.
-4. Submit short answer-only request; verify result within 60 seconds.
+4. Submit a short answer-only request; verify no time-based abort and visible manual **Continue in full session** after completion.
 5. Submit read-only inspection request; answer any external-read permission prompt inline.
 6. Submit mutation or multi-step request; verify reason and **Continue in full session**.
 7. Continue; verify normal session receives handoff and follow-up model.
-8. Close Quick Pane; reopen and verify quick transcript is absent.
-9. Select unavailable model or induce startup failure; verify inline recovery options.
+8. Close Quick Pane and reopen within three minutes; verify idle runtime, draft/catalog, and full transcript remain. Repeat after more than three idle minutes; verify fresh chat/runtime appears.
+9. Press `Ctrl+N` while idle and active; verify fresh empty chat in both cases and active work aborts before replacement.
+10. Verify pane opens at 640×360, grows only to 520px tall as transcript grows, then keeps transcript/model results scrolling internally.
+11. Induce settings-startup or model-catalog failure; verify visible **Retry**, disabled model/Send controls until recovery, and successful retry. Verify empty catalog stays blocked without unsupported actions.
 
 ## Exit criteria