Commit Graph
1740 Commits
Author SHA1 Message Date
Codeman maintainer b7c18ab5b7 Merge pull request #600: fix(mobile): shell Paste inserts a multi-line block as one bracketed paste 2026-10-13 18:27:11 +02:00
Codeman maintainer 70e03b8ae7 fix(preview): relative path links take m4v and ogv like rooted ones (#604/#606 integration)
#606 added m4v and ogv to the rooted file-path link pattern for the new video
uploads; #604's relative path pattern copies that extension list, and its
test pins the two equal. The relative pattern now carries them too, so a
relative `clip.m4v` links like `/home/u/clip.m4v`.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:26:46 +02:00
Codeman maintainer de06cb12d1 fix(i18n): drop the duplicate zh-CN Attach key (#606 merge-time fix)
ZH_CN already had `Attach: '附加'` (the tile grid's and remote discovery's
Attach buttons). #606 added a second `Attach: '添加附件'` later in the same
object literal, which won and turned every Attach into "add attachment".
The duplicate is removed; 附加 covers the phone toolbar button too.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:25:49 +02:00
Codeman maintainer 6c7d63d45e Merge pull request #606: feat(mobile): phone Attach button and camera-roll video uploads
Conflicts resolved against #605 (and #602's test mock):
- CLAUDE.md Prompt uploads: #606's rewrite is kept and #605's bounded
  workspace probe sentence follows the remote (SSH) refusal.
- test/routes/session-routes.test.ts: the remote-pane-history and
  bounded-path-probe mocks are kept beside #606's _resetPasteRateBuckets
  import, and both PRs' paste-image tests are kept.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:21:52 +02:00
Codeman maintainer 3d341f46a4 fix(i18n): zh-CN for the working folder row and its copy toasts (#593 merge-time fix)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:19:17 +02:00
Codeman maintainer f0b762e745 fix(session-options): label a remote or Docker session's working folder correctly (#593 merge-time fix)
The row said the path is "on the machine hosting Codeman" for every session.
A remote SSH session's workingDir is the path on the remote host, and a
Docker session's CLI runs inside the container the host folder is mounted
into. openSessionOptions now sets the description from session.remote /
session.docker (the id it targets was added with the layout fix).

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:19:17 +02:00
Codeman maintainer 771c23e261 fix(session-options): stack the working folder row so long paths fit (#593 merge-time fix)
The row was a plain flex `.set-row` with a non-shrinking actions column, so
the <code> took its full max-content width: the label column collapsed to
0px, the path overlapped the description and the Copy button left the row
(off-screen on a phone). It now uses the stacked `set-row-block` layout the
Session color row already uses, and the path shrinks and wraps.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:19:06 +02:00
Codeman maintainer 0ff9b798af Merge pull request #593: feat(session-options): show the session's working folder with a copy button 2026-10-13 18:19:01 +02:00
Codeman maintainer 33c6cf5d20 chore(gesture): revert the stray gesture bundle path comment (#591 merge-time fix)
#591 rebuilt the committed gesture bundle in a worktree whose node_modules sat
in a sibling directory, so its esbuild source comment named that local path.
Nothing in the PR touches the gesture source; the bundle goes back to master's
bytes.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:18:22 +02:00
Codeman maintainer 9b9752d997 Merge pull request #591: fix(sessions): restore a session's colour, pin and image watcher after a restart 2026-10-13 18:18:22 +02:00
Codeman maintainer 836402c2c4 fix(terminal): a replay strips the 8-bit C1 form of OSC 52 too (#611 merge-time fix)
xterm.js still parses OSC 52 introduced by C1 OSC (\x9d) and ended by C1 ST
(\x9c). OSC52_STRIP_RE matched only the 7-bit forms, so a direct-PTY session
that emitted the C1 form re-copied old text on every replay.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:17:18 +02:00
Codeman maintainer f073742cf7 fix(terminal): an OSC 52 copy hands focus back to the terminal (#611 merge-time fix)
When navigator.clipboard is missing (plain HTTP, install.sh's LAN option) or
refuses the write (an unfocused window), _copyText falls back to a hidden
textarea whose select() takes focus, and removing it left focus on <body>:
the next keystrokes after Claude Code's /copy went nowhere. The handler now
restores the element focused before the copy, as Auto Copy does.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:16:56 +02:00
Codeman maintainer f3d7a86c16 Merge pull request #611: feat(terminal): copy OSC 52 clipboard writes from the PTY into the browser clipboard 2026-10-13 18:16:20 +02:00
Codeman maintainer 34b867025d fix(tabs): Escape from a flat-rail web-tab menu returns focus to the row (#582 merge-time fix)
#582 made the web-tab actions menu reachable on the flat rail (ContextMenu or
Shift+F10 on a role=tab row), but closeTabGroupMenu only looked for a tree
item or a Focus shortcut, so Escape dropped focus to <body>. A `.session-tab`
trigger now counts too.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:15:21 +02:00
Codeman maintainer 5d31e42db0 fix(tabs): Follow parent again keeps a Focus pin (#528/#582 integration)
#528's browser followParent operation rebuilt the target ref as
{ kind, id } to drop its manual placement, which also dropped #582's
`focus` flag. The browser PUTs whole layouts, so the pin was erased on the
next save. The operation now copies the ref and removes only `placement`.
#528's row-menu test also expects #582's 'Add to Focus' entry now.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:15:02 +02:00
Codeman maintainer a6ce52d3e7 Merge pull request #582: feat(tabs): pin tabs to a Focus section on the vertical rail
Conflicts resolved against #528:
- CLAUDE.md Owner tab layouts: #582's Focus sentences and #528's
  tabGroupId / spawnedTabsFollowParent sentences are both kept.
- i18n.js: both features' ZH_CN entries kept.
- tab-layout-browser.js: the section-model JSDoc documents #528's
  `options.parents` and #582's `focusRefs` return field.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:13:03 +02:00
Codeman maintainer a38eab49cd Merge pull request #528: feat(tabs): place new, child and web tabs into groups
Conflict resolved in docs/api-reference.md: #607's prompt cache section and
#528's tabGroupId section are both kept.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:10:37 +02:00
Codeman maintainer 9c56f7823d fix(preview): a shell prompt's trailing $ or # stays out of its folder link (#604 merge-time fix)
The folder link pattern treats `:` as a boundary and allows `$`/`#` in a path,
so every default bash prompt (`user@host:~/proj$`, `root@c:/workspace#`)
became a folder link to a path that does not exist, and clicking it showed
"Folder not found". folderLinkPath now strips them, so the link opens the
prompt's cwd.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:09:08 +02:00
Codeman maintainer 90034d07d7 Merge pull request #604: feat(preview): open ~/, relative and hard-wrapped paths; folder grid 2026-10-13 18:08:12 +02:00
Codeman maintainer 62227145dc Merge pull request #607: feat(telemetry): read out Claude Code's prompt cache state per session 2026-10-13 18:07:15 +02:00
Codeman maintainer 2e65daff83 style(terminal): re-wrap the _maybePageCliTranscript JSDoc to 120 columns (#527 merge-time fix)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:06:35 +02:00
Codeman maintainer 0e11012470 Merge pull request #527: fix(terminal): page a hollow Codex transcript with PageUp/PageDown from the first wheel event
Conflict resolved in terminal-tile.js's @dependency header: #601's
SHELL_LOAD_BASE_QUERY/SHELL_LOAD_QUERY and #527's pageKeysForGesture and
app.tabAlerts are all listed.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 18:06:24 +02:00
Codeman maintainer 78ed6bf004 docs(terminal): only a full=1 pass without alt=recording is never mux-visible (#601 merge-time fix)
The geometry-retry comment in selectSession and the #435 invariant said the
route never answers full=1 with mux-visible; a shell load's alt=recording on
an alternate-screen pane now is. The conclusion holds (a useFullHistory pass
never sends it), so only the stated reason changes. Also lists
SHELL_LOAD_BASE_QUERY in terminal-tile.js's @dependency header.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:59:20 +02:00
Codeman maintainer 36f8afdaba Merge pull request #601: fix(terminal): load a shell from tmux's rendered history, not the byte recording
Conflicts resolved against #583 and #588:
- One history-depth field: #583's `historyLines` / `capturedHistoryLines`
  (reported as `paneHistoryLines`) is kept, and #601's lines cut now reads it
  instead of a second `capturedHistorySize`. #601's `alternateOn` /
  `capturedAlternateOn` joins it, and the one display-message format string
  ends `#{history_size} #{alternate_on}`.
- selectSession keeps #588's full=1 switch for panes with no scrollback, gated
  off remote and Docker sessions, and #601's SHELL_LOAD_QUERY for shells.
- computeHistoryTruncationNotice names a line count when paneHistoryLines is
  known (#583) and otherwise uses #601's remainder wording.
- docs: api-reference lists both historyLinesCut and paneHistoryLines; the
  notice invariant keys on history_size while the shell's automatic load is
  the one place alternate_on decides.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:59:07 +02:00
Codeman maintainer 1855d9423b fix(terminal): remote and Docker tabs keep the byte tail on a switch (#588 merge-time fix)
A remote (ssh) or Docker session's local pane hosts a nested tmux client in
the alternate screen, so it always reports paneHistoryLines 0 while its byte
tail carries the real scrolled-off output. The new full=1 switch therefore
replaced their scrollback with the visible frame on every revisit. Those
sessions now keep the tail; the pinned test string and the invariants note
say so.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:49:43 +02:00
Codeman maintainer 7ba1e51c51 Merge pull request #588: perf(terminal): tab switch to a fullscreen Claude tab loads only its current screen
Conflict resolved in config/test-suites.ts: both new browser test globs kept.
2026-10-13 17:49:16 +02:00
Codeman maintainer ee038ad83c fix(terminal): the history notice waits for the pull already in flight (#583 merge-time fix)
_smoothScrollBy and the touch paths call _maybeLoadMoreHistoryOnScroll on
every frame. Once the viewport sat at row 0, the next frame hit the real
pull's in-flight early return, so its reveal ran at once and the notice
showed the pre-pull tail state for the whole pull. A later frame now leaves
the reveal to the call that started the pull.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:48:11 +02:00
Codeman maintainer f2d5af2b4c Merge pull request #583: fix(terminal): partial-history notice only when there is history to load 2026-10-13 17:47:37 +02:00
Codeman maintainer 50a1b58d2b Merge pull request #613: Give the codeman agent skill to every agent CLI, not just Claude Code 2026-10-13 17:42:58 +02:00
Codeman maintainer 9e8a851a2c fix(spreadsheet-preview): filter and index a shared worksheet rels list once (#612 merge-time fix)
ExcelJS's worksheet pattern is unanchored, so several parsed worksheets can
share one sheetNo and therefore one rels list. repairExcelJsModel filtered and
indexed that list once per worksheet, which let a crafted admitted file run
past the 20 s preview timeout. Each distinct list is now filtered and indexed
once per sheetNo, and the relationship Type strings are built once.
SPREADSHEET_ASSET_VERSION is refreshed for the edited core.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:42:52 +02:00
Codeman maintainer 79b9181932 feat(file-preview): open markdown clicked in the terminal in the rendered viewer
An in-workspace .md path clicked in the terminal went to the log viewer
(tail -f, a "streaming" header over raw text), because md was kept out
of FILE_PREVIEW_EXTENSIONS on purpose; every other surface already
rendered it. Owner decision 2026-10-13: markdown is a document to read,
so md and markdown join FILE_PREVIEW_EXTENSIONS and a click or tap opens
the file preview, rendered by default with the MD pill for source. The
main pane, every grid tile and the split's Pane B share the link
provider, so all of them route the same way. Other in-workspace text
(.log, .txt, .json, code) keeps the log viewer and its live follow.

FILE_PATH_LINK_PATTERN now also links `.markdown` and `.htm`, which
were not clickable at all.

Tests: file-preview-markdown.test.ts now asserts md and markdown ARE
routed to the preview (and plain text is not); terminal-touch-tap pins a
tapped .md, .markdown and .html opening the preview while .log, .txt and
.json open the log viewer, and a tile-registered provider routing the
same way against the tile's own session. Docs: CLAUDE.md, the
architecture invariants and the wiki (Working With Files, Mobile Guide).

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:19:02 +02:00
Codeman maintainer 8f62faf130 refactor(file-preview): one rendersAsPage() for the terminal and the preview
What counts as an HTML page lived in three places: HTML_VIEW_EXTENSIONS
on the server, a /\.html?$/ regex in the terminal link provider and a
HTML_PAGE_EXTS set in panels-ui.js. The two frontend copies are now one
rendersAsPage() next to previewsInFileViewer() in constants.js, used by
the link provider (and through it every grid tile and the split's Pane
B, which share registerFilePathLinkProvider) and by openFilePreview()
and its toolbar. A new test pins it equal to the server's list.

toggleFilePreviewPage() now returns the preview load, so a caller (and
the coming frontend test) can await the flip.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:14:53 +02:00
Codeman maintainer 7dd78d9d71 fix(sw): leave rendered HTML pages to the network
The fetch handler skipped only /api/, so a popped-out /html-view/ page (a
top-level navigation the worker intercepts) was put into CacheStorage
despite its `Cache-Control: no-store`, and the offline fallback could
serve it after logout had revoked the capability. /html-view/ is now
skipped like /api/. The new test runs the shipped sw.js against a stub
worker scope and drives its real fetch listener.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:13:29 +02:00
Codeman maintainer 2797098aa5 Merge pull request #547: feat(file-preview): render local HTML files as sandboxed pages
Conflicts resolved:
- CLAUDE.md: kept master's Prompt uploads paragraph and its longer
  File-path links paragraph (XLSX admission limits), and inserted the
  PR's html-view sentence after the download-only rule.
- src/web/public/terminal-ui.js: kept master's tile-aware
  getSessionId() in the shared file-path link provider and added the
  PR's .html routing to the file preview on top of it.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 17:03:00 +02:00
Zoltan Zombory 4ecf7fa228 fix(terminal): strip OSC 52 from replays and copy each clipboard write once
xterm parses write() data after chunkedTerminalWrite has already cleared
_isLoadingBuffer, so the main terminal's replay gate never held: a tab switch
or reconnect re-copied the newest OSC 52 in the replayed history. Strip every
OSC 52 from the replay beside the DEC 2026 markers (OSC52_STRIP_RE). Tiles
keep their own _bufferLoading gate, which holds there.

One Claude Code copy arrives up to three times (its own OSC 52, tmux
load-buffer -w, set-clipboard forwarding), so identical text within
OSC52_DEDUPE_MS is dropped. The manual-copy fallback opens only in the
focused tab, so a background tab on the same session stays quiet.

The replay tests now drive the real chunkedTerminalWrite into
@xterm/headless and fail without the strip. Test harnesses that supply
DEC_SYNC_STRIP_RE by hand get OSC52_STRIP_RE too, and CLAUDE.md records
the rules next to Auto Copy.
2026-10-13 13:54:15 +02:00
Codeman maintainer 59b92f2991 fix(spreadsheet-preview): open openpyxl workbooks that have a comment, table or chart
The XLSX preview hands an admitted copy of the workbook to ExcelJS 4.4.0,
which links each worksheet to its comments, comment notes (VML), tables and
drawing by looking the relationship Target up verbatim in maps keyed by its
own spelling, then dereferences the result unchecked. openpyxl, the library
most agents write .xlsx files with, uses absolute Targets
(/xl/tables/table1.xml), its own comment part name (xl/comments/comment1.xml,
which ExcelJS never loads) and a drawing XML ExcelJS parses to nothing. One
comment, table or chart was enough for "Spreadsheet preview failed".

repairExcelJsModel() (spreadsheet-xlsx-core.js) runs on ExcelJS's parsed
model right before its reconcile: it re-points a link to the key ExcelJS
stored for the same part (so an openpyxl table is kept) and drops a link,
tablePart, drawing or hyperlink reference that would dereference nothing.
It adds nothing and resolves keys as own properties only. The worker wraps
workbook.xlsx.reconcile per instance to call it, and throws if a future
ExcelJS renames that method. SPREADSHEET_ASSET_VERSION is refreshed.

Tests: unit tests on model objects, and three real openpyxl 3.1.5
workbooks in test/fixtures/xlsx that fail the worker tests without the hook.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 12:37:36 +02:00
Codeman maintainer ef071ed485 feat(agent-skill): give the codeman skill to every agent CLI, not just Claude
The agent skill only reached Claude Code: injection wrote
.claude/skills/codeman, which no other CLI reads, and the per-session
preamble was seeded for claude alone. A Codex, Gemini, OpenCode,
Antigravity, Grok, DeepSeek or OMP session inside Codeman never learned it
could start and drive workers.

- Registry field capabilities.agentSkillDir (a closed enum, since
  clis.json can set it and it decides where the server writes) names the
  skills directory each CLI scans. Claude keeps .claude/skills; those seven
  CLIs get the same skill in <case>/.agents/skills/codeman (each measured to
  scan it; Claude Code does not, so it never sees two copies). pi gets none:
  a project skill makes pi ask "Trust project folder?" at startup.
- The preamble is seeded for every local session whose CLI takes the
  skill; nothing in it is claude-specific.
- The skill text is CLI-neutral and names the two CLI-side failures to
  report instead of retrying: a sandbox that blocks loopback (Codex's
  default), and commands that do not carry this session's environment
  (Codex 0.162 runs them in a shared app-server daemon).
- POST /api/sessions injects after the launch pre-flight, so a CLI that
  cannot start leaves nothing in the workspace.
- codeman skill install gains --mode; uninstall --case removes the copy
  from every directory the registry can write.
- The settings toggle moves to Agents & CLIs > All agent CLIs.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-13 12:33:47 +02:00
timkjrandClaude Opus 5.5 683a1c218f fix(mobile): send a shell Paste image to the session the dialog was opened for
The Paste dialog's image branch called _uploadAndInsertImages() without a
session, so it fell back to the active session at pick time: after a tab
switch with the dialog open, the image went to the other tab, while the
text Send already used the session captured at open.

Also corrects the comments around the shared bracketed write: the size
refusal now lives in _sendBracketedPaste(), the budget covers the shell
Paste too, and a shell without bracketed paste (dash, bash before 5.1)
gets the markers stripped by tmux, so every line but the last still runs
there.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-12 13:47:14 -05:00
Zoltan Zombory 435a56be10 feat(terminal): copy OSC 52 clipboard writes from the PTY into the browser clipboard
Programs inside the terminal (Claude Code /copy, tmux yanks with
set-clipboard on, vim) set the clipboard with OSC 52. xterm.js registers no
handler for 52, so on a headless host (no xclip, no DISPLAY) those copies
were silently dropped and never reached the remote browser.

Register one handler on the main terminal and on every TerminalTile (split
Pane B, tile grid). It decodes the base64 payload as UTF-8 and copies it
through _copyText, falling back to the manual-copy dialog when the browser
blocks the write. Read queries (`?`) are refused, and nothing is copied
during a buffer replay, so a tab switch or reconnect never re-copies old
OSC 52 sequences.
2026-10-12 20:39:07 +02:00
RandalixandClaude Sonnet 5.5 6bfc1b88d8 fix(preview): never glue the entries of a path listing into one path
`git ls-files`, `find` and `ls -R` print one path per row. An extensionless
entry followed by an equal-or-shorter one looked like a path broken short of
the edge, so `src/components` + `src/index.ts` linked `src/componentssrc/index.ts`
and the real link was gone. Measured with the shipped terminalPathLine over this
repo: `git ls-files` 8 false joins (one over 6 rows), `find src docs scripts` 5.

A next row is now a new entry when it opens with `./` or `../`, or has a `/` and
opens on the same top directory the broken token did. A wrapped path continues
inside a name, so the captures/2 + 026-10-10/... case still joins. After the
change: 1 and 0 (`docker/server.Dockerfile` + `docs/SPEEDRUN.md`, a broken token
with no extension that no row-shape rule can tell from a wrap).

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-12 12:05:56 +02:00
DDT f5e59283f9 fix(approvals): treat DeepSeek permission prompts as terminal-only alerts
Supervised-hook CLIs (DeepSeek Harness) raise Approvals Inbox items whose
answer route refuses keystrokes, but push notifications and the mobile/desktop
UI still offered Approve/Deny. Mark those items answerInTerminal, omit
approvalId (and therefore push actions) for supervised hooks, and show
"Answer in the terminal" / Open session instead of broken buttons.

Fixes #598 (functional part; docs gaps left for a follow-up).
2026-10-12 16:16:16 +08:00
JD ecb2e3d8f3 fix(telemetry): stamp a compaction on entering the null window
Claude Code's tracker nulls recache_tokens_if_cold at a compaction or a tool-result clearing and only counts the expected rebuild at the next request, which also closes the window, so no report carried both and the old rule never fired. Stamp on the first warm null report after one that carried a value, carry it through idle re-reports and drop it at the next request; the readout now says "Context compacted at HH:MM" since a clearing opens the same window.

Also persist a changed report so a crash restores the latest one, rename the session field to statusLineExporter (statusLineTelemetry is the registry capability) and document its values, word the stake as re-cached, move the readout block above the SSE staleness heading, and drop keep-warm mentions until it lands.
2026-10-11 12:36:59 -04:00
RandalixandClaude Opus 5.5 3857fba065 fix(preview): folder grid review — multi-user scope, bounded probing, smaller items
- Multi-user: a non-admin's folder must resolve inside their own userSpacePath
  or the session's workingDir (folderAllowedForUser, the path picker's rule),
  checked on the resolved folder before it is read and again on each served
  file after symlink resolution; 403 otherwise.
- listFolder probes the local folder with probePathKind({ pastCap: true })
  first (unknown -> 503 via describeUnknownPath, never 404), then async
  fs.realpath; resolveFolderFile uses probePathKind instead of fs.stat.
- The guard runs on the realpath before readdir.
- FOLDER_PATH_LINK_PATTERN ends a folder at ':' (quadratic `://` rescan on
  colon runs; PATH=/a:/b linked as one folder).
- Load counters drop stale preview / folder-grid responses; closing the grid
  invalidates a listing still in flight.
- validFolderFileName refuses a leading dot; the remote listing matches
  extensions in any case.
- Docs: folder grid in the z-index lists; multi-user rule and bounded probe in
  the invariants and security docs.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-11 14:03:20 +02:00
timkjrandClaude Opus 5.5 063e6fc076 fix(terminal): keep the byte-recording load for a shell in the alternate screen, report a lines cut only below the ceiling
Review follow-ups for loading a shell from tmux's rendered history.

tmux adds no history while a pane is in the alternate screen, and a
nested tmux client (remote SSH and Docker cases, or tmux run by hand)
holds its pane there, so a full capture of it is one screen. The
display-message query that reads #{history_size} now also reads
#{alternate_on} (PaneCaptureOptions.capturedAlternateOn), and a request
carrying `alt=recording` (only SHELL_LOAD_QUERY sends it) is answered
the old way on such a pane: the visible frame over the tailed byte
recording. Agents, scroll pulls and Load full history are unchanged.
alternate_on cannot tell a nested tmux from vim or less, so a local
shell with one of those open takes the same fallback while it is open.

A full capture without `lines=` was reported as a 'tail' cut whenever
the pane held more than the configured ceiling (possible on tmux before
3.7), offering a Load full history that could never return more. The
cut is now reported only for a client bound below the ceiling.

The response carries `historyLinesCut` when that bound left lines in
tmux, and the banner then says earlier output may be retained instead
of naming the bytes a later tail cut, which undercount what tmux holds.
A bounded grid tile's shell load sends its own `lines=` (via
SHELL_LOAD_BASE_QUERY). CLAUDE.md, docs/architecture-invariants.md and
docs/api-reference.md describe the shell load, the cut report and the
fallback.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 23:29:19 -05:00
Codeman maintainer 764910baca fix(web): hold the WS pill's queued-bytes suffix until a backlog persists
A paste, an IME commit or a fast word sits above the 4-byte hint threshold for a few
milliseconds until its ACK lands, and each one widened the header pill (`WS · 12B queued`) and
moved the gauges beside it. The suffix now appears only after the backlog has stayed above the
threshold for 1.2s; a drain forgets the burst and a timer re-renders when the hold elapses.
2026-10-11 11:37:09 +08:00
Aamer Akhter d6c87bbd27 fix(terminal): end the page-key debt lifetime once travel repays it
pageKeysForGesture kept `prepaid` set after the travel had fully repaid a
pre-paid page, so the one-second debt lifetime still applied: on a 38-row
pane, 19 lines of -1 at 32 ms left { pending: 0, prepaid: true } and a -1
300 ms later sent nothing. `prepaid` now clears as soon as the pending
travel reaches or crosses zero, so the leftover is plain travel and the
150 ms gap applies again. Touch contacts are unaffected (never split by
time). The mixed parity step's reversal grows from 20 to 25 rows because
the -5 rows left after repayment are now plain travel and cancel against it.
2026-10-10 22:52:15 -04:00
Aamer Akhter 3ea1e03aaf fix(terminal): one touch contact is one paging gesture; bound slow ticks and wobble
Review of #527 (Ark0N, 2026-10-11):

1. A slow touch drag paged a full screen per line. touchmove calls the
   fallback once per whole cell, so a slow drag's calls were further apart
   than the 150 ms gesture gap and each pre-paid a page (12 lines at 170 ms
   sent 12 PageUps). The touch paths now tag every call, touchmove and
   momentum frames alike, with a per-contact touchId (incremented on
   touchstart), and pageKeysForGesture never splits a contact by time; a
   new contact is a new gesture.
2. A pre-paid debt now survives up to PAGE_KEY_DEBT_MS (1 s) of quiet when
   the next event goes the same way, so slow small wheel ticks (40 px every
   250 ms, a Firefox 1-line notch) pay it back instead of each pre-paying.
   A reversal inside the gap (or inside one contact) drops the debt and
   accumulates without pre-paying, so a +/-3 px wobble no longer pages
   back and forth. A reversal after the gap still pages at once. Main pane
   and tiles share the helper, so both get it.
3. A wheel consumed by the dialog check now logs page-keys-dialog-open.
4. The Dashboard wiki names newer Codex beside OpenCode; the invariants
   paragraph describes the new gesture boundaries.

Tests: unit regressions through the real _maybePageCliTranscript, helper
cases, tile/main parity for ticks and wobble, and a Chromium test that
drives the real initTerminal touch listeners (12 PageUps before, 1 after).
2026-10-10 22:50:06 -04:00
Aamer AkhterandClaude Opus 5.5 afab34608a Merge origin/master (1.41.0) into pr/grouped-rail-placement
One conflict, in i18n.js SKIP_SELECTOR: master added '.tab-name' itself
(the same fix this branch made for zh-CN tab names), so master's line is
kept and this branch's duplicate comment dropped.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 18:19:01 -04:00
Aamer AkhterandClaude Opus 5.5 eaf924a291 Merge origin/master (1.41.0) into pr/codex-hollow-paging
Conflicts resolved by keeping both sides:
- terminal-ui.js: master's wheelDeltaWholeLines/sgrWheelReports next to
  this branch's pageKeyRows/pageKeysForGesture.
- terminal-tile.js: master's tile SGR wheel forwarding and Shift+wheel
  local scroll (_shiftScrollPending) kept; tile paging stays on the
  shared pageKeysForGesture state (_pageKeyGesture) with its per-tile
  dialog check.
- architecture-invariants.md: this branch's hollow-paging paragraph,
  with master's tile SGR forwarding and Shift+wheel sentences replacing
  the old "do not forward SGR wheel" note.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 18:18:02 -04:00
RandalixandClaude Opus 5.5 1010d80b4f feat(preview): folder grid — browse every previewable file of a folder
Folder paths in the terminal and the response viewer are now links
(rooted or ~/ without an extension, relative with a trailing slash); a
click opens a grid of the folder's supported files, newest first, with
subfolder chips and a parent button. A tile opens the file preview,
which pages through the folder with ‹ › / ←/→; the preview of any file
gets a "Show folder" button.

Server: POST /api/sessions/:id/folders grants + lists a folder (local
fs or one ssh round trip for a remote case) under the attachment guard;
GET …/folders/:folderId/files/:name/(raw|preview|thumbnail) serves one
file by bare name through resolveServableAttachmentPath, so browser
requests still never carry a path. Grants live in their own small
registry so a big folder cannot evict the session's attachments. The
attachment raw route's body is now the shared serveServableRaw helper.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-10 21:54:37 +02:00