The renderer no longer holds the whole history. Startup sends queued
changes, lets Main reconcile leftover pending/running records against the
stored tasks (same localized interruptedOnRestart line), then loads the
first page (200). The Activity page loads a full visible batch (500, as
before paging); "load more" shows the next batch and loads the next page
behind it; a filter switch refetches that filter's first page; leaving the
page trims to one batch. Counts, project/conversation groups, titles,
conversation status and the clear count come from activityHistory.summary
(refreshed on entry, by the page timer, the refresh button and after load
more / filter changes).
Records that are not loaded are still updated incrementally: the update
contract gains content-matched changes (upsert-call, update-request,
resolve-approval, settle-request, remove-request, remove-call) that Main
applies in SQL to shown rows; the store applies the same semantics to its
pages through the shared reference. update returns Main's stored call
records so a call whose earlier record was not loaded keeps one id. Task
refresh reconciles via Main SQL. Legacy localStorage records migrate as
incremental changes; the renderer no longer calls replace.
Tests: randomized sequences of every action, page loads, trims, flushes and
injected failures against the former full-list code on a real database;
first page <= 200 after startup; flush payload depends only on changed
records (including unloaded ones); ActivityPanel output with pages + summary
equals the full-list rendering for randomized datasets with duplicates;
restart interrupts leftover records with the same detail; clear empties the
database, pages and summary; SQL change semantics equal the reference.
Group settings tabs and forms in a single surface with consistent spacing. Rename the model section to cover model selection and execution limits, and use Default model for the default connection option.
Release note: 修复监督者设置页签与表单贴边的问题，统一设置布局，并将默认模型选项名称改为默认模型。
Schema 57 rebuilds activity_history_records with order_seq (newer = larger;
a move to the front assigns a new maximum) and stored generated columns
(record_id, occurrence, status, status_group, kind, request_id,
conversation_id, project_id, created_at) with indexes. The order is migrated
from record_order_json exactly; legacy duplicates keep their occurrence and
only the first is shown. record_order_json is no longer written.
ActivityHistoryRepository owns the SQL: page({before, limit <= 500, filter,
projectId, conversationId}) newest first by index, summary() with the counts
and conversation titles/status the Activity page derived from the full list,
reconcile() joined with tasks with a renderer-localized detail line, update()
applied directly in SQL without reading the history, replace() for the
legacy migration, clear(). New activity-history:page/summary/reconcile IPC
with zod validation; page and summary run on the readonly worker.
perf:main measures 24,000 records: startup load 106 ms on Main (14 MB) ->
first page + summary 13.5 ms on the worker (0.6 ms loop block, 114 KB).
The renderer no longer sends the whole history on every change. Domain
actions on an external activity store update the list and queue the matching
contract changes; activity-sync.ts flushes them every 250 ms via
activityHistory.update (<= 20,000 per call), keeps failed changes queued in
order with backoff retry, and flushes on quit and unmount. replace remains
only for the legacy localStorage migration; clearing uses the new
activityHistory.clear IPC. App's activity state is gone; ActivityPanel reads
a selector. update is now required in the desktop API.
48d7526 widened the options dialog first-focus selector to inputs, so the new story graph switch above the menus took initial focus. Keyboard users no longer landed on the expert / work mode menus like the model picker. Restore the button-only target; the switch stays reachable with Tab.
Add a supervisor text-model selector using existing connections and a per-conversation story graph switch above knowledge settings. Hide the switch when Supervisor is disabled and enforce conversation preferences at local and remote tool execution boundaries.
Release note: Choose a dedicated supervisor model and control whether each conversation can query existing stories and experiences. Enabling graph access does not start a review or automatically insert context.
Main module evaluation drops from ~420 ms to ~330 ms. Types stay as import type; the DeepSeek Harness chunks stay static because the protocol chunk shares code with Main startup paths.
Runtime creation, MCP launches, Obsidian, native clients and the tool-environment IPC await the initialization promise instead of startup awaiting it before the window path. A failure still fails startup with the same diagnostic.
Pause scroll writes while the jump-to-bottom smooth scroll runs; keep rows with pending questions, approvals, recent interactions and selection ends mounted; cache Mermaid results; copy text spanning unmounted rows; anchor the sidebar on id changes; disable native overflow-anchor for windowed lists; write terminal output on the leading edge and before clear; forward remote events every 16 ms; flush streaming text before copy/export; add performance principle P10 (no UX regressions).
Default the notes list to 200px with a 160px minimum and hide it automatically only below 600px. Keep the detail toolbar toggle available to open a drawer, with dismissal and note-switching support. Restore reading positions after rich text content effects populate the record.
Include the existing document-table and file-tree scroll anchoring styles in this file-level commit. Validation: 143 related tests, Electron layout regression, typecheck and changed-file lint passed; repository lint is blocked by unrelated App.tsx changes.
Release note: The notes list takes less space and remains accessible through a drawer in narrow windows. Existing saved list widths are preserved.
Document shared review timelines, stories and experiences, spiral navigation, conversation performance work, and Execute configuration behavior. Include schema 56 backup and cost boundaries, Agent tool-schema compatibility, and four successful current-source real Host calls with checkpoint persistence. Full release validation runs in CI.
Persistence and history loading, refresh scheduling and the agent event handler work against the conversation store outside React; App.tsx drops to 9,566 lines and the lint ceiling to 9,600.
The chat pane decides re-windowing once per frame from scrollTop and cached row heights instead of reading layout per scroll event. Rendered Markdown hast trees are cached, so remounted rows skip parsing. Sidebar lists of 60+ conversations mount only rows near the viewport. The perf driver scrolls the windowed sidebar to reach a row.
conversations:list-summaries and conversations:get run on the readonly query worker. Activity history replace validates and serializes only changed records; a new activity-history:update IPC writes one record. Adds npm run perf:main.
ESLint allowlists (only to shrink) for DatabaseSync, synchronous fs in src/main and direct window.goodbuddy subscriptions in components, plus a max-lines ratchet for App.tsx; docs demos and vendored scripts are no longer linted. perf:check (or GB_PERF_CHECK=1 perf:app) fails a non-throttled benchmark that exceeds build/perf-thresholds.json. Records the results in the PERF plan.
Remote semantic events were written with one or two synchronous SQLite transactions each before being forwarded; recoverable conversations also rewrote the whole assistant message per event. Events now commit together at each remote checkpoint (before the Agent can acknowledge), and before non-remote, question and terminal events, on a size cap or a 50 ms timer. Only newly stored events are forwarded, in order; a failed batch rolls back and blocks later commits. Applies to interactive, background and recovery runs.
The first message batch now grows by 20 per frame after the first paint instead of one large commit, keeping the reader at the bottom or in place. The cached pane order is reconciled into state by a pure helper instead of reading and writing a ref during render (react-hooks/refs).
Each open of the time spiral created a WebGL context that was never released, so repeated use hit the browser limit and lost the oldest context. The view now releases it on close, shows a local retry on creation failure or context loss, and the lazy chunk retries once before an in-canvas error.
Leaf batches retry at half and a quarter of the text budget, navigation merges retry once; provider errors are not retried. Adds the long-session acceptance script and records the results.
Opening a long conversation got faster because of the two-step pane render (09c415b), not the live message store, and its catch-up frame is now a long frame. Records which concurrent changes follow the batch rules, and adds Story Graph search and batched remote event writes to PERF-15/16 and the list-order state to PERF-14.
Runs the review on an isolated copy of a real database with DeepSeek, injects one provider failure, resumes, and checks coverage and zero-call reruns. Credentials come from an env file passed on the command line.
A freshly mounted pane first renders the trailing 20 messages and the rest of its 80-message batch right after the first paint, keeping the bottom or the reader's place. Saved scroll positions and note targets still render the full batch at once. Keeps the write-on-change guard for --chat-context-height.
Search matches query terms (CJK pairs) instead of only the whole phrase, previews show readable fields first, and the tool description says when to call it and when to stop.
Configuration apply previously required a separate native confirmation even in Agent Runtime Execute mode, and failed in clients without an approval callback. Use existing work-mode and tool authorization instead, preserving Ask read-only access and request-scoped plan validation. Update regression tests, tool descriptions and documentation.
Validation: 267 focused tests passed; typecheck and changed-file lint passed. Full tests timed out with failures in other modules; repository lint reported errors in unchanged files.
Release note: Execute 模式下通过 Agent Runtime 修改 GoodBuddy 配置不再额外弹出确认框，Ask 模式仍保持只读。
Add App regression coverage and document five-second sidebar sorting, live row updates, and deferred reordering during interaction. The ordering hook and App wiring were already included in ee4eb73. Focused validation: 23 tests passed.
Also includes pending workspace changes: unsaved-change guard, keep-alive cache and live message store, knowledge workspace dialogs and perf driver updates.
perf:app gains GB_PERF_CPU_THROTTLE (CDP renderer throttling) and a keep-alive memory scenario with post-GC heap samples; perf:knowledge times synchronous FTS, vector and hybrid search at 5k/20k/50k chunks. Results and conclusions recorded in the performance plan (PERF-05, PERF-09, PERF-12).
Each source version is extracted once for every review scope; events carry spans and projects from their sources, and attention density is counted from messages. After publication, events are assigned to project features and sub-threads (cross-project stories only when enabled), with user rename, merge, remove, move and undo. The story graph gains a lazily loaded Three.js time-spiral view. Includes the 3D concept demo, real-data DeepSeek check scripts and docs.
Wait for detail navigation and continued editing focus, and focus the cancel button before synthetic clicks so confirmation restores the actual opener. Keep draft discard and final entry focus assertions unchanged. Validate in CI without touching parallel work.
Isolate Obsidian tests from the default browser capability, check the additional image-tool card, and center native overlay click targets below the frosted header. Retain existing permissions, identity, hit-testing and keyboard assertions; preserve unrelated uncommitted work.
Release committed image-save tools, note editing improvements, and enabled-by-default browser and frosted appearance preferences. Include bounded Linux Host save evidence and bilingual Desktop/Agent notes. Preserve all pre-existing uncommitted design and article files; full validation and packaging run in CI.
- Add a Cancel action to the new-entry composer with discard confirmation
- Remove the 880px cap so note content matches the toolbar width
- Focus the chat composer only after a deferred workspace leave commits
- Return focus to a sensible target after cancelling new/edited entries and switching entry type
Expose save_image across direct model, MCP gateway, DeepSeek Harness and remote Agents (written on the remote host via chunked reads), and show the image MCP tool list in settings.
Update the Electron probe for the two-line shortcut placeholder after greeting removal. Preserve all prefix-equivalence checks and give only the exhaustive Markdown corpus test 15 seconds. Validate through CI per user instruction.
Scroll settings controls into the unobscured area below the measured sticky header, retaining native interaction and hit-testing assertions. Give the full legacy outbox migration fixture a scoped 15-second timeout. Product code and bilingual release notes are unchanged; validation remains in CI.
Document streaming rendering improvements, on-demand activity refresh, local custom MCP failure isolation, managed model source selection, and opt-in device sharing. Agent remains 0.15.1 and schema remains 52. Per user instruction, validate the exact candidate through CI without additional local tests.
- Segment streaming Markdown so completed blocks are reused and only the growing tail is re-parsed; fall back to whole-document parsing for document-wide constructs
- Memoize conversation sidebar rows and cache per-conversation derived activity, ordering and pending approvals
- Activity page loads on entry, refreshes every 15 s while visible, and has a single header refresh button
- Include pending settings, model download source, device sharing and knowledge gateway changes in the working tree
- Add Electron streaming input benchmark and regression tests
Check detail navigation focus before the auto-focused creation input takes over. Preserve return-to-list and reading-position assertions, and include the reviewed sharing first-read test and independent validation record. Validation runs in CI per user request.
Include device-sharing metadata and the persistent New note entry in bilingual release notes, align five-app documentation, and update the configuration fixture. Per user instruction, validation proceeds through CI without more local tests. Agent remains 0.15.1.
Scope a 15-second budget to the 161-message switching scenario while retaining all render-count, DOM reuse, draft, attachment, and scroll assertions. The full App suite, typecheck, and lint pass.
The Linux full suite sampled a window-control color during its transition. Wait for CSS transitions to finish while retaining exact style, contrast, geometry, and native interaction assertions.
Add an opt-in persistent frosted glass setting for the top bar and runtime checklist, with chat scrolling behind the header while preserving scroll anchors and window controls.
Retry transient local MCP initialization failures once after cleanup, include redacted connection diagnostics in feedback, and index lifecycle events to reduce execution statistics scans.
Release note: Add optional frosted glass appearance with chat visible behind the title bar; improve local tool connection recovery and diagnostic feedback.
Make draft thumbnails and queued images open the shared image viewer without parsing. Keep queued attachment menus read-only and remove OCR source and settings entries while preserving explicit text extraction in drafts.
Validation: 14 focused tests, typecheck and lint passed. Full suite: 5534 passed, 3 failures outside this change in conversation scrolling and frosted-glass descriptions.
Release note: 草稿和队列附件中的图片现在支持点击放大查看，无需 OCR；队列只读菜单不再显示解析来源和解析设置。
Keep unchanged history panes and task strips from rerendering during typing, and stabilize message callbacks across cached conversation switches. Preserve streaming updates, scroll positions, drafts, and asynchronous attachment ownership.
Add App regressions for render isolation, warm switches, retry, and image recovery. Focused tests, typecheck, and lint passed; the full suite encountered unrelated settings failures.
Release note: 减少任务运行时输入和切换已缓存长会话的重复渲染，保留实时输出、滚动位置和会话草稿。
Align the notes list, detail toolbar and comment direction row, removing stacked top padding. Let text editing controls fill the record card while preserving text and action padding and canvas spacing.
Add Electron geometry regressions and document the spacing rules. Release note: Magic Notes now aligns pane headers and removes excess gaps around text editing toolbars and input areas.
Quick Notes could open the full workspace before its selected record was mounted, losing the scroll request. Perform entry navigation after React commits and retain a deterministic regression. Align the manual-compaction fixture with layered model selection and loaded conversation state.
Release note: 从快速笔记打开完整记录时，等待目标内容挂载后再定位，避免偶发只打开页面却未滚动到所选记录。
Include note tags and layout, workspace controls, and OpenCode launch/tool-name updates in bilingual notes. Update schema 51 guidance and align the Electron Supervisor regression with the implemented suggestions UI.
Keep the notes list beside the open note, compact the detail view with a full-width sticky toolbar, and make Back to overview return to a full-width all-notes grid. Add manual note tags with AND filtering, a tag manager for rename, merge and delete, schema 51 tag tables, and tag support in the note Agent tools.
Scroll the file list independently so view selection, refresh, creation actions and directory navigation remain visible. Keep the selected Files entry available in projects without Git.
Add component coverage and 16 Electron layout scenarios for themes, window sizes and scroll retention. Focused tests, typecheck and lint passed; the full test run timed out after 200 seconds.
Release note: Workspace file controls remain visible while scrolling, and projects without Git retain the Files entry.
Use gbd-/gbc- prefixes with a 6-char conversation hash instead of 20-char per-request hashes, lease per-conversation slots so parallel requests never share a name, and expose gateway custom tools as <server hash>_<name> within the 64-char provider limit.
Every 'Open in terminal' launch for OpenCode copied the bundled plugin dependency tree (about 3,600 files, 52 MB) into a temporary directory, adding 2.2-2.4 s on Windows before the TUI started. The tree is now copied once into terminal/opencode-config under user data and reused until its bundled version marker changes. Skills, Ask-mode plugins and model configuration are still generated per launch and cleaned up on close. OpenCode's own TUI startup (about 3 s) is unchanged; the remote Agent path is unaffected.
Release note: 在终端中打开 OpenCode 时不再每次复制插件依赖，第二次及以后的打开速度明显加快。
Return text and canvas records to read-only content after a successful save. Automatic analysis no longer reopens the editor; failed saves preserve drafts for retry. Update regression coverage and UI behavior documentation.
Validation: 118 MagicNotesWorkspace tests, typecheck and lint passed. Full test suite exceeded the 120-second timeout.
Release note: 笔记中的文字和画布记录保存修改成功后自动退出编辑状态，自动评论完成后不再重新打开编辑器；保存失败时保留草稿以便重试。
Heartbeat no longer writes its own report; it triggers the shared incremental review and, per plan intervention, derives evidence-backed suggestions from the review's graph changes. Manual review is incremental by default with explicit reanalysis. The Supervisor tab is renamed Smart heartbeat with heartbeat plans, and the redundant inner plan frame is removed.
Render checklist details in an anchored floating panel with bounded scrolling, outside dismissal and Escape focus restoration. Reuse the shared overlay mechanism and update layout documentation and regression coverage.
Validation: 25 focused tests, Electron checklist layout and keyboard scenarios, and lint passed. Full-suite and typecheck blockers outside this change were reported separately.
Release note: Runtime execution checklists now expand in a floating panel without pushing conversation content down.
- Build runtime history per turn: a cancelled or failed reply keeps its partial
text plus an interruption marker (or a marker-only placeholder), so its user
message is never left unanswered and merged into the next request.
- Use the same history for send, manual compaction and Main compact validation.
- OpenCode: keep a cancelled prompt in the reused native session and prefix the
next turn with a one-time synthetic interruption notice; failed prompts are
still removed.
- Add an opt-in DeepSeek live test covering direct model and OpenCode.
Keep the Runtime icon and label clear after a conversation-specific selection. Preserve override details and the reset action inside the picker, with regression coverage and bilingual release notes.
Release note: 移除会话 Runtime 按钮图标与名称之间的圆点，保留菜单内的单独选择提示与恢复项目默认操作。
Wait for the enabled send control rather than its initial presence. Preserve the question rounds, saved answers, skip, and reload assertions; production behavior and bilingual release notes are unchanged.
mesalogo