The round-robin walk capped what it listed, not what it read: every queued
folder was read whole, sorted and held until drained, so one GET could pull
millions of entries (100 folders of 50,000 files killed a 400 MB heap). A read
now keeps only the entries the listing can use, and MAX_WALK_ENTRIES_READ stops
opening folders at 250,000 entries, leaving the rest queued and flagged.
A folder also spent its whole share on its own files, leaving nothing for the
subfolders it had just found, alone in its round or beside a sibling, so search
hits inside them were lost. Its files now count as one more share beside those
subfolders. Math.max(directoriesInRound, 2) would have fixed only the sole-folder
case and cut a wide root's folder list short.
Folders below the walk root are probed before their readdir (#516), since the
walk reaches folders the depth-first one never did.
Filtering at the read also clears the false flag on a folder whose only unread
entries were hidden or excluded, and an exactly-full workspace now finishes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With the folder field empty, Create New's Browse opens in the cases folder, and New
folder there answered "Use Add Case > New Case". There is no tab of that name (the
label is Create New), and a Create New user is already on it. The refusal now reads
right from Link Existing and from Create New's custom folder. Its test pin, the code
comment, the manual, api-reference, CLAUDE.md and architecture-invariants move with it.
The block() helper in the New folder picker test returned an empty string for a name
that is not in session-ui.js, so .not.toContain passed and a renamed picker dropped out
of the check. It now asserts the name was found.
The manual's path picker list said two places and left out Create New, which the New
folder paragraph names. It now lists every Browse that opens the picker, and that
paragraph is rewrapped to the page's width.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Applies the review of #766.
The live lookup ran tmux on the bare PATH of the ssh command. On a host
where tmux sits in ~/.local/bin or /opt/homebrew/bin it found no pane, the
2>/dev/null hid that, and the panel stayed on the launch conversation after
a /clear. It now goes through withRemoteTmuxPath(), inside the sh script and
scoped to the $(...), so the other tools keep the PATH they had. The fake
tmux of the test block moved to $HOME/.local/bin for the same reason: that
directory is first on the new PATH, so a real tmux is never asked.
The two ssh commands get exec. Node signals the local sh -c, and dash stays
the parent of a plain command, so ssh survived the timeout. A test with a
fake ssh shows the process dies now and survives without exec.
The comment at the meta budget claimed a meta past it comes with a later
read. A meta comes only with a subagent's first read, so it is never sent.
Docs: the store is named as the exception to "never copied to local disk",
remoteProbePaths and the remote transcript lookup are described, and the
line saying a remote session loads no store is gone.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The fair rounds give a directory its node before reading it, so every folder
still queued when the 5,000-node budget ends sat in the tree with empty or
partial children and read as an empty folder. Keep the node on the tree cursor
and set an additive incomplete: true on every cursor left in the queue, then
draw one dim, inert row under such a folder while it is expanded. The envelope
check ignores keys it does not know, so old and new clients stay compatible.
Stop each round at the top once the budget is spent, before the shift: the
rest stay queued unread, and flagged, rather than read and sorted for nothing.
Say what the walk is: a per-directory round-robin (breadth-first), fair per
directory and not per subtree, with search results now shallow-first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The API reference still said a registry failure on create and clone is
500 and did not list the rename's new 409s. CLAUDE.md said both files
are restored after a failed registry write, and left out the
config-backup restore exception.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The security wording claimed a folder strictly below workingDir was the whole
argument, which the symlink fix showed was only the lookup's half. The rest is
docs that were missing: liveWorktree, root=live and worktree in the API
reference, the mechanism section the CLAUDE.md paragraph had no pointer to, what
a root=live query matches against, and the limits a reader meets in the wiki
(the Git panel stays on the session's folder, a session started below the
repository's top is not followed, a page under .claude/worktrees opens as
source).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The edit history panel now covers remote claude sessions. The sync stays
path-based; its deps get the session and server.ts routes them to the local
reader or to two new ssh reads in remote-files.ts:
- remoteLocateClaudeTranscript: one round trip; the conversation the remote
pane's claude writes now (~/.claude/sessions/<pid>.json via the pane's
process tree, since remote hosts send no hooks), else claudeSessionId,
else the session id, newest $HOME/.claude/projects/*/<id>.jsonl.
- remoteReadClaudeEditSources: one round trip per read pass, with the local
reader's semantics (offset clamp, block end at offset+len, new files read
their tail, shared main-first budget, onlyRequested, 8 KiB meta cap).
So /clear catch-up, "Load older", cursors and the store LRU work remotely
unchanged. Reads use buildRemoteFileCommand + runWithRemoteSshLimit only:
a sleeping host is a connect timeout -> unavailable, stored entries kept.
Tests: both scripts run on a real sh against a temp HOME; the read is
compared byte for byte with readClaudeEditSources on 60 seeded random
transcripts. A transitive import guard pins that the edit-history read path
reaches neither remote-wake.ts nor session-routes.ts. Browser: a remote
session lists images without thumbnails, a click previews. Docs updated.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Consolidates the four contributor changesets, and rewords the em-dashes this release
added in comments and docs. The five in test/settings-save-failure-ui.test.ts stay:
they assert existing toast text.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The startup hint is no longer typed, a Discord reply is one of the input paths that
set fromUser, edit-history is a state file, POST /api/filesystem/mkdir is in the API
reference, and the store's header says which routes really refuse it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Owner decision. Typing the note into a brand-new non-Claude tab spent a model
turn per tab and, for a spawned worker or a scheduled run, landed with the first
real prompt. Claude keeps its hint through the system prompt, and every other CLI
is told nothing. The setting goes with the path it gated: sessionIdle, the fresh
set, isHintableSession and the session:idle wiring. sendInput loses `fromUser`,
which was only ever false for the hint, and the deps test pins that a reply is a
person's own input.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`cwd` beside the fixed `workingDir` reads ambiguously on a public API surface, and
`paneCwd` says where the value comes from, like `paneExit`. Also pin the field on
the payload the browser really receives (getSessionStateWithRespawn), and give the
pane-directory text its own architecture-invariants paragraph.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Round-four review smaller items: api-reference now states the opposite-of-apiKey behaviour for an omitted defaultContextLength, and the wiki sentence follows the code, modelContextLengths[id] ?? defaultContextLength: any stored figure beats the default.
- share one context-length resolution (discovered figure, endpoint default fallback) across the routes and boot recovery, with a recovery test
- record listed per-slot context figures with a slot count of 1 and force the slot-count fallback once per rediscovery
- open up the codex model launch token for qwen3-style ids (model-plain-colon), pinned by a spawn-golden case
- pass the full discovered window to opencode as limit.context
- translate the picker search strings and document the discovery precedence
A direct clone whose git-slot wait timed out came back with no spawnError, so
the cleanup emptied a folder git never touched. GitRun now says whether git
started and the cleanup requires it. Cloning into a folder at or inside a
Docker case workspace is refused, as git status, fetch and worktrees do, since
git would run the new repository's hooks and filters on the host.
A direct clone into an empty folder emptied the folder after any failure, which
deleted a concurrent attempt's finished clone when git had written nothing.
Clones into one folder are now mutually exclusive in-process, the cleanup is
skipped when git refused the folder, never ran, or the mount is unresponsive,
and a name linked while the clone ran is a 409 that keeps the folder.
Documents path for POST /api/cases/clone.
- lone digit rule in CLAUDE.md, invariants, user docs, wiki, API reference and code comments
- say which installs have the codeman command
- reload() comment back above reload(), merge the sendInput comments, typedHint in the hint comment
- drop the stray dot after --urgent in the startup hint
- config-backups wiki: name the discord.json token, not the URL
- comma in the CLAUDE.md auto-named list
- codeman notify exits 1 when the route answers 200 with data.ok false
- notify request times out at 30 s instead of hanging
- fileProblem probes the file with the bounded probe before realpath
- the thread map records its parent channel; a new channel starts afresh
- threadFor returns its failure instead of sharing this.last
- discord-notify: per-session token bucket (6 a minute), 429 RATE_LIMITED, plain line from codeman notify
- fileProblem probes the working folder with the bounded probe before realpath
- sendTest uses its own DiscordApi and leaves this.cfg alone
- drop the resumed-tab check from the typed hint (only Claude's field was read)
- first sessionRenamed per session records the name without a PATCH
- boot goes through ensureLoaded, so one gateway starts
- a thread created without an id reports the failure
Claude keeps the silent system-prompt hint (startupHint, on). Other CLIs are sent the typed message only with typedHint, which costs a model turn per tab.
Adds the codeman notify command, a startup hint (Claude via --append-system-prompt, other CLIs once on first idle), files saved from thread attachments into the tab folder, and --file uploads limited to the session folder.
Alerts for permission prompts, questions, idle and errors post into a per-tab thread; a reply from an allowlisted user is typed into that tab. Thread creation, thread cleanup on tab delete and pinging are configurable. POST /api/sessions/:id/discord-notify lets a CLI ask for the user. Setup is in docs/discord-bridge.md.
tmux reads a pane's directory from the process cwd, which is always the
resolved path, while workingDir keeps the spelling the case was created
with. For a workspace reached through a symlink every reading looked
like it was outside the workspace, so the browser sent each relative
link through the attachment route instead of the workspace routes.
Each session now resolves workingDir once in the background, bounded by
probePathKind() and a timed realpath(), and setCwd() writes a reading at
or under the resolved path back in workingDir's spelling. Until that
settles only a reading lexically inside workingDir is taken. An unknown
probe or a failed realpath leaves readings untranslated and retries
after WORKSPACE_REALPATH_RETRY_MS.
Also corrects two comments: cwd is written to state.json but never read
back, and a blank pane path is usually a pane whose program exited.
applyPaneExits() only applied the manager's directory when it returned a
value, so after a successful read a session that was split into several
panes, or was no longer on the socket, kept its stale cwd. The manager
already keeps its map across a failed or empty read, so the server now
applies its answer as given, undefined included, and setCwd() broadcasts
the clear once.
A blank #{pane_current_path} on a single-pane session (tmux could not read
the process cwd that tick) now carries the previous entry forward in
derivePaneCurrentPaths(), so only a split or unlisted session clears.
The server tests now drive the real TmuxManager with only the tmux read
faked. docs/api-reference.md documents SessionState.cwd.
The merged PRs (mostly #714's git fetch) added about 120 em-dashes in
comments, docs, test names and three settings strings. Each one is now a
comma, colon, semicolon or parentheses. The English settings strings in
index.html and their i18n.js keys change together, and the zh-CN value
for the fetch interval uses full-width parentheses instead of a double
dash. The one left is the regex in constants.js that matches a dash
typed as a word, which is behaviour, not punctuation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The self-gate now says the skill reads saved Codex conversations offline outside a session; the staging
reference says only tagged input gets 409 during a guarded submit (the rest is held), and the API
reference notes a managed write past a full hold queue falls back to a direct write. Also: missing
reference rows, agent --help naming inspect/stage/submit, the tunnel opt-in in Remote Access, MCP wording.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
readDockerCases() reads an unreadable or half-written docker-cases.json as [], so a fetch (which runs
repository hooks, and runs unattended on the interval) could run inside a Docker workspace. The fetch
paths now read it through a strict reader: the Fetch button answers a failed run (state 'error') and the
scheduler skips the tick, undoes its picks and logs once. The status and diff routes are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflicts resolved by keeping both sides:
- CLAUDE.md (Agent CLI): #694's opening (an HTTP client, the experimental
agent-session-routes behind agentCoordinationEnabled) with #708's clause
on the directory spawn and its two refusals.
- docs/api-reference.md (codeman agent): #708's spawn paragraph and --json
shape, with #694's inspect/stage/submit in the verb list and the guarded
endpoints named before "adds no route for the other verbs".
- src/cli-agent.ts (@fileoverview): #708's quick-start / directory spawn
routes and #694's experimental routes in one endpoint list; the two
node:fs/promises imports (#708's realpath+stat, #694's open+stat) merged
into one.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflict in CLAUDE.md (Adding Features, App setting) resolved by keeping
both sides: #714's sentence that a setting the server acts on must be
synced, after the `.strict()` clause, and #719's mutateSettingsFile() rule
at the end of the bullet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflict in src/web/routes/system-routes.ts (PUT /api/settings) resolved by
keeping both sides: #718's line that never stores an unauthorized
tunnelEnabled:true runs first, then #719's serialized mutateSettingsFile()
write replaces the inline read and plain writeFile.
test/routes/system-routes-tunnel-guard.test.ts: #718's new tests asserted
plain writeFile calls on settings.json. With #719 every write goes to a temp
file renamed over it, so the assertions read writes through a settingsWrites()
helper (writeFile contents keyed by their rename), and the concurrent-stop
harness applies a write and the stop at the rename.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- src/web/routes/session-routes.ts:2102: the guarded-submit 409 now applies to tagged input (clientId + seq) only; untagged input is held like a WebSocket frame. The raw write (:2228) goes through holdForGuardedLease and answers 200 once held (a send-and-wait reads `delivered` from the held write; a full queue releases the waiter and answers 409); a mux write takes its own turn in writeViaMux.
- test/guarded-lease-held-input.test.ts: real-Session route tests for an untagged raw POST during the lease (written after the Enter) and a held send-and-wait (delivered: true); both fail without the fix.
- test/routes/session-routes.test.ts + test/mocks/mock-session.ts: the 409 test now covers tagged input only; new tests for untagged raw (held), untagged mux (own turn) and a full hold queue (409); MockSession gains holdForGuardedLease/releaseGuardedLease and a lease-aware writeViaMux.
- src/session.ts:4787 (guardedInputHeld JSDoc) + write() + docs/architecture-invariants.md: write() refuses during the lease only, so cron's paste mode can land ahead of input still draining.
- docs/api-reference.md: the input route's 409 is for tagged input; untagged input is held.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- src/web/settings-writer.ts:110: export assertSettingsWritable(), the writer's own read-for-write rule run on its own
- src/web/routes/case-routes.ts:1860: an admin's rename checks settings.json before it ends any session or moves the folder, and answers 409 CONFLICT with the writer's message (placed before the session-starting check, so nothing is awaited between it and the lock)
- test/routes/case-rename-routes.test.ts:314: the torn-file pin now expects 409 CONFLICT (no stoppedSessions, like the other pre-checks)
- test/routes/case-rename-routes.test.ts:331: new test, a live session with stopSessions and a torn file: 409 CONFLICT, cleanupSession not called
- docs/api-reference.md: the Settings paragraph says an admin's rename answers this refusal as 409 CONFLICT before ending sessions
- src/hooks-config.ts:1161, src/prompt-cache-keepwarm.ts:341: comments describe the atomic writer, not an in-place rewrite
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- docs/api-reference.md:1222: the tunnel status shape is { running, url, loginConfigured } (getStatus() returns only running and url)
- src/web/routes/system-routes.ts:1233: the unauthenticated-tunnel warning keys on !isLoginConfigured(), so it no longer fires in multi-user mode
- test/install-sh-invariants.test.ts:537: install.sh must not contain --allow-unauthenticated-tunnel either (it writes service argv as well as env)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The input lease is held across the awaited tmux Enter, so WebSocket frames
keep arriving while it is held. A tagged frame (seq N) was refused and its
seq rolled back, but the browser's WebSocket path does not wait for ACKs:
seq N+1 landed after the lease and moved the watermark past N, and the
redelivered N was then dropped as a duplicate. The keystroke was lost.
Input that arrives during the lease is now held and written in arrival order
when the lease is released, whatever the submit's outcome. A WebSocket frame
is applied and ACKed only once written; managed writes (writeViaMux) and
named keys (send-key) wait for their turn the same way, and a held write that
is still being sent holds everything behind it. A keystroke held before the
guard runs still moves the input revision, so it still invalidates the Enter.
The hold is capped per lease (256 inputs, 1 MiB): past it the WebSocket is
closed with 1013 and nothing later from it is applied, so the browser resends
every unACKed frame in order, and send-key answers 409 CONFLICT instead of
logging a tmux failure as INTERNAL_ERROR. POST .../input keeps its 409 for the
lease plus the held input's drain; its browser drainer is serialized.
Docs: correct the api-reference paragraph that said these keystrokes are
redelivered, and describe the capture fallback and the per-scope cursor cap
from the previous two commits in api-reference, architecture-invariants and
the changeset.
`codeman agent spawn` always sends its directory as an explicit
workingDir, so the server cannot apply its no-workingDir rule and
installs workspace hooks (and an injected skill) into ~/.claude when the
caller's cwd is HOME. A bare spawn whose cwd resolves to the home
directory now refuses with exit 4 before any request and names
`--working-dir ~`; an explicit `--working-dir ~` is still sent. Paths
are compared by realpath, so a symlinked HOME matches, and an
unresolvable HOME or cwd falls back to the resolved path without
throwing.
The directory-spawn --json shape now reads the same in the README, the
API reference, the wiki page and the changeset: `ready` when a
readiness wait ran, `readinessError` when that call failed, and
`startupError` when the start failed. A test pins all four.
The network acknowledgement that install.sh writes for every passwordless LAN install also satisfied the tunnel guard, so one click on the welcome screen's tunnel link published full terminal control to a public URL with no login and no confirm. The tunnel now needs CODEMAN_ALLOW_UNAUTHENTICATED_TUNNEL or --allow-unauthenticated-tunnel (carried by web -d and service install), and with no login the browser asks before every enable, reading a new loginConfigured field from GET /api/tunnel/status.
The guard is now decided once per request and required again where the tunnel starts, so a save that found the tunnel running can no longer restart it unguarded after a concurrent stop.
Closes#706
Quick-start saved lastUsedCase with an un-awaited, non-atomic read-modify-write, and the settings PUT, model config, case order, case rename, Ralph start and the boot tunnel reset each ran their own. Two overlapping truncate-then-write calls left one write's bytes over the tail of the other (the CI flake in test/quick-start.test.ts), and most writers then read the torn file as {} and wrote back only its own keys, erasing the rest of the user's settings.
All of them now go through one writer that runs one change at a time, replaces the file atomically, and refuses (409 CONFLICT) to overwrite a file that does not parse.
`execFile` signals the `git` parent only, so a fetch that times out can leave its http helper
still running, and on the interval that repeats every run with nobody watching. `runFetch` now
spawns `detached`, so the fetch and the helper it spawns share one process group, and the kill
is SIGTERM to the negative pid then SIGKILL `FETCH_KILL_ESCALATION_MS` later; the escalation is
not cancelled when the promise settles, because the caller is answered at the timeout while the
group is still dying. The output cap is kept, and the clone's slot queue is not needed here.
The tick is two passes: every enabled local session is collected by folder, a folder is due when
ANY session open on it is due, the folder is probed once for the group, and the run is published
to every session on it. A single pass decided by the order the sessions were read, so [long,
short] published only the short one and the 3600 s session's row never updated again while its
folder was fetched every 30 s.
`gitNonInteractiveEnv` writes BatchMode ssh as a default, not an override: `core.sshCommand` is
read inside the same slot and the same bound as the fetch it serves, and `fetchEnvFor` keeps the
repository's answer when it has one, because that is usually the setup that works on a machine
with several ssh setups. The `accept-new` host key it can write is stated next to the BatchMode
note: a prompt is stopped, a write is not.
The write claim is scoped to the Git panel and to the timer, with cloning and worktree management
named as the writes that happen when the user asked for them, and the zh keys sit with the group
of the UI surface their English string belongs to.
Tests: the process-group test proves the tree by group membership, because the helper is a
grandchild (`git fetch` runs `git remote-http`, which spawns `git-remote-http`) and because the
group is what the negative pid reaches, and it remembers each process once found instead of
re-scanning a group it had already emptied. The non-zero-exit test changed premises: `git fetch
--quiet` in a repository with no remote exits 0 on git 2.47.3, so resolving empty stdout there was
correct, and the case that has to reject is a remote pointing at a path that is not a repository.
The join test asserts after the run and does not await the request before the held run is
released, which would wait on the run the test is holding.
Stage now re-reads the pane after typing (every 50 ms, up to 1 s) and
issues a receipt only for a frame with the same generation, the input
revision its own write moved, and a composer hash equal to the
message's. Codex repaints 30 to 60 ms after the keys, so the first
capture was often stale and stage refused with the text already typed.
Every refusal after typing now says the message may be in the composer
and points at the Ctrl+U recovery. A visible-frame capture reads the
screen and the caret in one tmux invocation, with the old two calls as
the fallback.
Codex draws its suggestion popups above the composer and keeps the
footer, so a staged /stat verified and submit ran /status. Stage now
refuses a leading / or ! and a last word starting with @ or $ before
typing, and stage and submit refuse a frame with popup rows just above
the composer. The pattern and characters are registry data
(composerPopupLine, composerCommandPrefixes, composerCompletionSigils),
read from the codex 0.162.0 bottom_pane sources.
Submit waits until 250 ms after the staged keys: codex turns an Enter
within 120 ms of a typed burst into a newline. POST .../input answers
409 with its seq returned while a guarded submit holds the input lease,
instead of 200 for bytes it never wrote.
Also align the attribution memo comment with the code (a verified file
whose incremental scan fails is not re-walked, since the full walk
reads a superset), and list stage as a fromUser source in CLAUDE.md.
Codeman maintainer