…one prefix from fold headers

* feat(tui): improve message scrolling and rewind picker

* feat(tui): add double-escape rewind shortcut

Squashed PR lsdefine#393: update README.md and add installation docs.

…e support

- app.js: triple-guard for CJK IME composition (compositionstart/end + keyCode 229)
- lib.rs + Cargo.toml: add tauri-plugin-single-instance to prevent zombie processes
- tauri.conf.json: add "dmg" target and png icons for macOS bundling

# Conflicts:
#	frontends/desktop/src-tauri/Cargo.lock
#	frontends/desktop/src-tauri/src/lib.rs

… merge PR lsdefine#395

…evtools by default

…-turn outputs

… instead of _turn_parts

…l check in loop

… tips

Adds to frontends/tuiapp_v2.py:
- /rename <name>: rename current session; collision-checked across in-memory
  sessions and the persistent registry. Topbar and sidebar refresh in place.
- /continue <name>: resume by stored name in addition to the existing index
  form (/continue 1). The picker (/continue with no arg) labels each entry
  with its stored name when available.
- /cost [all]: per-session token usage (input / output / cache_create /
  cache_read), elapsed time, request count, and context-window %-left. The
  "all" form lists every tracked thread, including ended sessions.
- ask_user picker upgrades: free-text escape hatch on every list, Esc-back
  navigation, single/multi-select switch from a [多选]/[multi] question hint,
  candidate sanitizer that handles JSON-envelope debris.
- Runtime tip footer (`└ Tip: ...`) that rotates short usage hints.
- Topbar status chip showing the active model + idle/busy state.
- Cross-platform launcher hardening: auto-install of rich/textual on first
  run, and a hint when running under git-bash/mintty.

New modules:
- frontends/session_names.py: JSON sidecar at temp/model_responses/
  session_names.json keyed by log-file basename; touched only by /rename and
  /continue <name>. continue_cmd is not modified.
- frontends/cost_tracker.py: per-thread TokenStats, captured via
  monkey-patches on llmcore._record_usage and llmcore.print (Anthropic SSE's
  final output count only arrives via the latter). Idempotent install().

No core files touched.

…ine#397)

feat(tui): /rename, /cost, /continue <name>, ask_user picker, runtime…

Adds Ctrl+T theme cycling across six palettes: a custom ga-default
matching the existing github-dark scheme plus Textual built-ins
(nord, gruvbox, dracula, tokyo-night, textual-light). Themes flow
through Textual's native register_theme + get_css_variables, so CSS
references $ga-bg / $ga-fg / etc. that resolve from the active theme;
ga-default uses our exact palette while built-ins derive the missing
slots (border/dim/muted/border_hi) from resolved bg/fg via a blend.

watch_theme keeps the C_* module globals (Rich Text colors) and the
role-color dict in sync with the active theme, invalidates message
caches, and remounts the current session. Rich Markdown rendering
also passes a per-theme RichTheme so code blocks, links, headings,
and quotes follow the active palette instead of Rich's frozen
defaults. text-muted / text-disabled are deliberately not inherited
from Textual: those resolve to 'auto NN%', a Textual-only syntax
Rich Text rejects.

…sdefine#399)

Rephrased docstring in cost_tracker.py to clarify API-mode tracking semantics.

Display a concise startup banner with model name when running in interactive TTY mode.

The five topbar chips (session name / model / effort / tasks / clock)
were hardcoded github-dark hex values introduced in the upstream
topbar redesign, so they stayed cyan/amber/lavender on every theme.
Add five chip_* slots to _DEFAULT_PALETTE and derive them for
built-in themes from primary / secondary / warning / accent / success,
keeping five distinguishable hues across themes. watch_theme syncs
the C_CHIP_* globals on switch.

Also move the '▾ fold' indicator from the LEFT column to the RIGHT,
just before the clock. The left column packs identity + session +
running status pill into a ratio=1 ellipsis column, so a long status
('running 2m 15s') was eating the trailing fold glyph as '...'.

The clock chip was github-dark's brighter success green (#56d364),
which read as more vibrant than the rest of the muted palette and
clashed with C_GREEN. Use the same #7ec27e the sidebar uses for the
active session marker, so all 'positive accent' greens in the chrome
sit at one saturation level.

These are agent-internal metadata: <summary> already feeds the sidebar
preview and the fold title, <thinking> is meant to be hidden. They were
leaking into the chat body, and worse, CommonMark parses
'<summary>X</summary>\n<body>' (no blank line between) as a single HTML
block that swallows the following body line — so the model's actual
reply disappeared from view whenever it wrote the summary tag immediately
before its answer with only a single newline.

Strip both tags in _render_md before passing to the Markdown parser.

Rich's divide_line treats a Chinese run as one indivisible word and
bumps it whole to the next line when it doesn't fit the remaining
space, wasting cells like 'AI ↩ 助手...'. Patch both rich._wrap and
textual.content so Text.wrap (assistant Markdown) and Visual.to_strips
(user messages) pack leading CJK chars into the current line's
remainder, then fold the rest at full width.

Non-CJK words keep stock behavior; code-block content remains
untouched because the patch only adjusts wrap break positions.

…ocs/

_render_md now produces two renders: the existing narrow ANSI roundtrip
for display (mouse selection still anchors via segment.style.meta) and a
wide render at width=10000 that contains only structural \n. An alignment
walk maps each visual line back to a position in the wide source, so
SelectableStatic.get_selection extracts source text without wrap-induced
newlines.

The aligner handles three layouts:
- 1:1 (no wrap): narrow line maps directly to wide line
- N:1 wrap: hanging indent on continuation lines is stripped, and the
  whitespace eaten at wrap points (e.g. spaces between English words) is
  recovered from the wide line
- centered single-line (Markdown headers): wide and narrow have different
  leading-pad widths; source uses the lstripped content, and narrow's pad
  becomes per-line indent that selection clamps over

The two CJK-aware drop-ins (divide_line for Rich/textual.content, and
compute_wrap_offsets for textual._wrap/_wrapped_document) shared
identical greedy-pack-then-fold core logic, written twice. Extract into
_fold_chunk_cells so each call site only handles interface shape (Rich
words vs Textual chunk regex, tab handling).

Also extend the patch to cover textual._wrap.compute_wrap_offsets and
textual.document._wrapped_document — the TextArea / Static-via-Content
paths use compute_wrap_offsets, not divide_line, so the prior patch
missed input-box and user-bubble wrap. Drop try/except defensiveness
around the Textual patch (we target latest Textual; let real breakage
surface), split _CJK_WRAP_RE into commented sub-ranges, prune WHAT
comments.

Verified: non-CJK output remains byte-equivalent to upstream across
both divide_line and compute_wrap_offsets fixture sets; live render at
width 60 packs the first line to ~98% utilization on input box, USER
bubble, and AGENT bubble.

…utput

extract_ui_messages now renders tool_use headers and tool_result fences
into the assistant content using the same string format that agent_loop
yields live, so fold_turns folds restored sessions identically to live
chat. Tool-only response rounds no longer drop the user message, and
Turn 1 also carries a marker so the first turn folds like the rest.

_user_text was misclassifying any prompt whose text block didn't start
with '### [WORKING MEMORY]' as a real user message — but auto-continue
prompts can interleave [SYSTEM TIPS], [System] regenerate triggers,
[DANGER] guard injections, and similar. Treat any prompt carrying a
tool_result block as auto-continue (the natural signal that this is
the next round of an in-flight LLM call), and extend the injection
marker list as a secondary guard for the rare tool_result-less
synthetic prompts.

…ine#599)

* feat(tui): /continue 恢复 plan 卡片 + Ctrl+S/Ctrl+C 路由修复

plan-card 恢复：plan 模式判定改用日志中结构化的 enter_plan_mode
tool_use（continue_cmd.find_plan_entry），不再扫聊天文本——用户打字
伪造不出。is_active/resolve_path 新增 restored_path 信号，AgentSession
新增 restored_plan_path 字段，plan_scan_baseline 退化为仅供 current_step
的 📌 扫描用。

Ctrl+S restore 卡死：_stash_cleanup_restore 由 self.text= 改为
clear()+_insert_via_keyboard，走编辑管线避免长会话全屏 relayout，与
stash 清空路径对称。

Ctrl+C 弹回收紧：_cmd_stop 仅在该轮未被消费（agent 尚未产出回复）时
才把 user 文本弹回输入框，避免已答轮次污染输入框 / 重发导致历史重复。

* feat(tui): plan 卡片可滚动（4 行任务窗口）+ TODO 渲染为 [ ]/[x]

planbar 拆为固定表头/步骤行（#planbar-head）+ 任务滚动区
（#planbar-tasks，max-height:4）。最多露 4 个 TODO，其余滚轮/PageUp
翻看，去掉旧的 5 行预算与 ⋮ +N more 截断；表头与当前步骤行固定不滚。

任务标记 ☐/✔ → [ ]/[x]，与 plan.md 源文件写法对齐。

The tool-cards change (lsdefine#593) tightened _md_line_has_box_drawing to exempt
pure-horizontal '─' runs as markdown hr. But a SIMPLE-box Markdown table's
only box glyph is its header rule, so that exemption dropped whole tables
out of the passthrough copy path and into the wide-render aligner, which
maps selection cell-columns to character indices — misaligning copy of any
table with CJK (wide) cells.

Remove the pure-horizontal exemption so a table's header rule is detected
again and the table run stays passthrough. Card gutters (└─ / │ / └) keep
their own prefix-based exemptions, so tool cards are unaffected. The only
cost is a real markdown hr copying as a dash run — the pre-lsdefine#593 behavior.

…owser (lsdefine#598)

web driver init nudges the browser by opening a page when no session
appears; os.startfile raises AttributeError on Linux/macOS, killing
the retry loop on headless servers. webbrowser.open is stdlib and
cross-platform with the same effect.

修复个人微信前端在 headless 容器内无法登录

* feat(tui): workspace 项目模式 + @ 文件引用补全（v2/v3）

- workspace: /workspace 设/off/picker、junction 复用 project_mode 记忆、
  per-session(v2)/进程级(v3)、/continue 恢复、session map 持久化绑定状态
- @ 补全(completion-only): 子序列模糊 + path-like 目录补全(~/ / ./ C:\)、
  提交期 @路径作为普通文本交 agent
- v3 filterable 焦点链 picker(输入框作焦点环对象、free_input)
- 新增共享模块 at_complete.py / workspace_cmd.py；核心零改动

* feat(tui): @ 路径绝对化 + 默认根=temp + 未绑显示完整路径

- 默认根 os.getcwd() → agent 工作目录 <GA根>/temp（与 file_read/code_run 一致、不随启动 cwd 飘）
- 提交期 @相对 → @绝对（display 保留相对短路径），让 agent file_read（相对自身 cwd）找得到
- 未绑 workspace 时候选显示完整路径（根不直观）；索引忽略 model_responses 会话日志噪音
- 启动预热 temp 索引 + candidates_for 惰性兜底（任何根首次访问自动建）

…lsdefine#606)

- macljqCtrl.py: macOS implementation of mouse/keyboard/screenshot/window
  enumeration mirroring ljqCtrl API, plus AX accessibility control tree
  (AXElements/AXFind/AXPress) as the macOS equivalent of UIA
- computer_use.md: add section 3 documenting the macOS platform branch
- .gitignore: whitelist memory/macljqCtrl.py

… prevent UI freeze

ESC cancel sent session/cancel RPC but never cleared the busy flag, leaving the UI stuck if the server-side cancelled event never arrived. Also clear busy on all sessions when the bridge closes so pending poll loops exit cleanly.

Co-authored-by: dilong888 <dilong888@users.noreply.github.com>

v3 的 _do_restore 此前只把历史载入内存 backend.history，不碰当前进程日志。native 后端逐轮只把新增消息 append 到 model_responses_{pid}.txt，恢复出的旧历史从不写入当前日志——续聊只追加增量，下次 /continue 扫到的本进程文件就只剩续聊部分，旧历史丢失。

对齐 v2 _do_continue_restore：恢复前 reset_conversation（快照+清空当前日志），恢复后 copyfile（源日志 → 当前日志），续聊便接在完整历史之后。

…d, AXClick)Feat/macos ax enhancements (lsdefine#611)

* feat(memory): enhance macOS AX control (resolve_pid fix, enabled field, AXClick)

- _resolve_pid: fix bundle_id branch (used un-imported AppKit), 3-tier match (bundle id > app name exact > substring)
- AXElements: collect 'enabled' field (SOP: check disabled before click)
- AXClick: AXPress-first then fallback to physical-coord Click, honest success判定 via pixel diff
- AXFind: add enabled_only filter, refactor if-forest into _hit helper

* docs(ljqCtrl): tighten macOS AX guidance, drop boilerplate & app-specific detail

- collapse 5-line cross-platform import boilerplate to one line
- generalize control-identifier tip (remove app-specific identifiers)

Annotate config block to guide new GA instances: enumerate candidate
mykey.py variable names and experimentally probe which config works;
only print var/field names, model, apibase host/path, status codes and
error types (never full dict or apikey/token); include a config format
example. Also flag that apibase/endpoint may differ per provider/proxy
at the two request-building sites.

…conflict) (lsdefine#617)

/continue N 默认改为「原地续」:接管原会话日志、后续轮次写回同一文件,取代旧的
镜像复制(supersedes b4356c7 的 mirror 方案),不再每次续接都 fork 新会话、日志增殖。

- 空闲会话 → 直接原地接管;被活进程占用 → 弹窗确认后才复制一份续。
- 每会话出生持锁(temp/model_responses/.locks/<logid>.lock),整进程共用一个心跳
  线程每 5s touch mtime、30s 无心跳判死可接管;atexit 干净释放,崩溃/强杀靠超时兜底。
- 切走/新对话不再「快照+清空」(原 _snapshot_current_log 因 logid≠pid 实为死代码,
  从未生成过快照),旧日志原样留作空闲会话,新对话铸新 logid。
- list_sessions 新增 exclude_log,修正 exclude_pid=getpid 失效导致当前会话出现在
  自己列表里的问题。
- continue_cmd 仅新增函数,reset_conversation/restore/handle 未改动 → 其他前端
  (IM/qt/streamlit 等)行为完全不变。
- 改动仅限 continue_cmd.py + tui_v3.py + tuiapp_v2.py;rewind/worldline 逻辑不在本 PR。

Update WeChat group 21 QR code image.

Co-authored-by: AspasZhang <AspasZhang@users.noreply.github.com>

…ne#621)

In-place /continue retargets agent.log_path to the restored file, so the
reset bind `_bind_workspace(None)` ran during restore was persisting
session_ws_set(path, "") — erasing that session's own workspace mapping to
"" (read back as "explicitly off") before we read it, which also short-
circuited the log-scan fallback. The continued session never re-entered its
workspace.

Add a `persist` flag to `_bind_workspace`; the restore-time reset passes
persist=False so it only refreshes in-memory state. The session→workspace
map is now written solely by explicit /workspace, /workspace off, and a
successful restore.

Ctrl+S (stash draft) cleared/restored the input via call_after_refresh,
deferring the visible state change to the Screen idle callback queue.
That queue is only flushed when the screen is layout/repaint-stable, so
under heavy streaming (multiple long-context sessions producing a near-
continuous stream of dirty regions) the deferred clear/restore could be
postponed long enough that the input never visibly updates while the
half-flipped _draft_stash flag leaves the box in an inconsistent state —
perceived as a freeze until the streaming settles.

The deferral only existed to keep the keystroke snappy back when reset()
rebuilt the TextArea document; reset() now routes through TextArea.clear()
(edit pipeline, no document rebuild), and the full clear/restore + resize
measures well under 1ms even on very long sessions. Running the cleanup
inline makes the keystroke authoritative: by the time the Key event
returns the buffer and stash flag are already consistent, independent of
whatever the streaming loop is doing.

Two related streaming-bridge defects found while reproducing:

- _on_stream referenced an undefined refresh_chrome in the exit-boundary
  replay branch, raising NameError; the exception propagated out of
  call_from_thread, so the done event never settled and the spinner spun
  forever. Unconditional chrome refresh now matches the main path.

- _consume_display_queue called call_from_thread directly; a raised
  callback re-raised via Future.result() and crashed the consume thread
  mid-task, again stranding the spinner. Wrapped in _call_stream so the
  consumer survives and done can still land.

Caveat: the user-reported complete freeze was not stably reproducible in
Textual Pilot under multi-session long-context streaming (the ~2s seen
there is Pilot._wait_for_screen waiting for widget queues to drain, not a
real main-loop stall). These changes remove the most plausible trigger —
the stash path dependency on async callback scheduling under load — but
cannot be claimed to definitively fix a symptom that did not reproduce.