killProcessOnPort() was a no-op on win32 (POSIX-only, via lsof/kill),
so a restart could leave the old OpenCode process holding the port
while a new instance spawned on a different one. That's a plausible
contributor to a chronic pattern seen in production logs: repeated
"OpenCode process exited, restarting" cycles and hundreds of
ECONNRESET/proxy errors over multiple days on Windows.
Give killProcessOnPort a real Windows branch: parse `netstat -ano`
for PIDs listening on the target port, filter out our own pid, and
force-kill each via `taskkill /PID <pid> /F` (no /T -- we don't own
that process, so only the listener itself is killed, not any
children it may have).
waitForPortRelease()'s existing soft-fail-and-warn behavior is left
untouched -- it's a deliberate safety net for any platform where the
port doesn't free up in time, not just Windows, and the restart
already rebinds event-stream readers to the actual resulting port via
onOpenCodeRestarted.
Creating a device key while the UI is open through a public https domain
(reverse proxy) dropped that domain from the QR payload whenever the
dialog passed a preferred LAN URL, leaving only the local IP and relay
as transports. Carry the non-loopback request origin as an additional
direct candidate (priority 20, between LAN and relay) so paired devices
can keep using the same domain on any network.
OpenCode Go now reads quota usage with a bearer API key from OpenCode auth.json
Removes the old workspace ID and browser cookie credential flow
Deletes legacy OpenCode Go credential files during upgrade
MCP authorization was broken in several stacked ways. The browser return
leg landed on the SPA behind the auth gate, so the system browser saw a
login page instead of finishing; the pending-context store silently
saved nothing because its route had no JSON body parser; and the
callback-URL config write started deferring behind Apply & Restart, so
authorization ran against a runtime without the URL and dead-ended on
OpenCode's loopback listener.
The return leg is now completed entirely server-side by an
unauthenticated GET /mcp/oauth/callback that only forwards a code whose
state matches a parked context. Desktop with the local server and VS
Code switch to OpenCode's native flow over its fixed loopback port —
no config writes or restarts at all, with a one-time cleanup of the
previously written callback URL — and its completion signal drives the
page instead of blind status polling. Remote, hosted-web, and mobile
keep the server-callback flow, applying a queued callback-URL write
immediately since authorization cannot wait for a manual restart.
Also: a server queued behind Apply & Restart now shows an Awaiting
restart badge and explanation instead of connect/reauthorize buttons
that can only fail, and Reauthorize is offered only while the server is
actually connected.
Relay demand now counts the authoritative transport signal: a request
arriving through the tunnel permanently marks the client usesRelay, and
hasActiveRelayClients also accepts lastTransport === 'relay'. Store read
failures no longer masquerade as no demand, so reconcile can't persist
enabled=false and sever paired devices on a transient error.
`getWorktrees` logs a warn-level line every time the managed OpenCode
process or any other caller passes a directory that is not inside a
git repository. The OpenChamber desktop main.log fills with hundreds
of these "Failed to list worktrees, returning empty list: fatal: not
a git repository ..." entries over a normal session.
The empty-list fallback is already correct (worktrees are an optional
feature), but the warning is noise that hides real git failures. Use
the existing `isNotGitRepositoryError` helper to suppress the warn
specifically for the "not a git repository" case and keep the
warning for genuine failures (lock contention, permission errors,
corrupt repos, etc.).
Adds a work-status panel beside the transcript. Context fill, model and
cost, todos, running subagents and the permission requests blocking
them, branch and working-tree state, MCP servers, pinned messages and
context sources were scattered across the header, the composer and the
context panel — a blocked subagent was reported nowhere at all. The
panel reads them from live channels rather than persisted history, and
becomes an overlay where the chat is too narrow to seat a column.
It is on by default, including for existing installs. Because it now
carries these readouts, the desktop header and composer drop the ones it
duplicates: todo and changed-files chips, usage and MCP tabs. VS Code
and mobile keep theirs — neither hosts the panel.
Fixes MCP authorization, which was broken from the panel, invalidated by
a directory switch through a redirect URI that encoded the working
directory, and left the desktop app in the background because browsers
will not follow a custom-protocol link without a user gesture. The
settings page no longer asks the user to understand the MCP spec before
adding a server: one field takes the command or the link, with the kind
inferred and a visible override, and client-registration fields appear
only when a server actually asks for its own credentials.
Also: skills load from the panel instead of only when the composer's
slash autocomplete opens; the header button names the current instance
rather than falling through to the word "Instance" for relay hosts.
Three new optional UI settings keys, all migrated. No change to stored
MCP server configuration.
Address the openchamber-ai review's non-blocking notes:
- Concurrency evidence: the torn-write test now injects a slow, chunked
writeFile (one open handle, file grows prefix->full) so a torn read is
deterministically observable in the 30ms window. A companion test runs the
naive direct writer under the same load and asserts torn reads ARE produced,
proving the atomicity test can actually fail on the pre-fix writer.
- Windows fallback comment: no longer claims the copyFile fallback is atomic;
it is called out as a last resort confined to Windows.
- Module map: document cli-settings-accessors.js in bin/lib/DOCUMENTATION.md.
The CLI's settings accessors wrote settings.json directly with writeFile and
read it leniently, with no strict-reader gate on relay identity. Running
'openchamber connect-url' while the desktop app is up could:
- tear the file for a concurrent reader in the app, tripping the relay
service's read and mapping it to {} (first-run);
- then regenerate the relay signing/encryption keys, changing serverId and
orphaning every paired device and push binding.
Move the accessors into a dedicated module that mirrors the settings
runtime's guarantees: atomic tmp+rename writes (with the Windows fallback)
so no reader can observe a partial file, and a strict reader that throws on
corrupt/unreadable payloads so identity regeneration is gated exactly like
the server runtime. Wire the strict reader into the CLI relay identity path.
Adds unit tests covering atomic writes under concurrent readers, strict-read
behavior, and that a corrupt settings file makes getRelayIdentity fail
instead of minting a replacement keypair.
When the project root is not itself a git repository, discover nested
repositories (depth- and visit-capped readdir walk via a new
/api/fs/git-dirs route), auto-select the first one, and show a repository
picker next to the branch dropdown to switch. Selections persist per
runtime and root; discovery failure is a distinct marker with a retry
action, never an empty success.
* fix(git): enable core.longpaths for worktree population
Worktrees live under a deep OpenCode data-dir path, so Windows checkouts
of deeply nested repos failed bootstrap with "Filename too long". Enable
Git core.longpaths before git reset --hard (web + VS Code) and surface
clearer path-length guidance when the filesystem still rejects a path.
Fixes#2746
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
* chore(vscode): keep ensureWorktreeLongpaths private
Avoid an unused export in the VS Code git service; the helper stays
local to populateWorktreeWithLockRecovery.
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
* fix(ui): update session-switch-resync test to current handleEvent/setSessionTodos signatures
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
* perf(build): split Shiki grammars/themes, CodeMirror legacy modes, and @pierre/diffs into on-demand chunks
Merging @shikijs/langs into one manual vendor chunk made the first language
request download every grammar (7.4 MB raw / 1 MB gzip). Letting Rollup split
these packages per dynamically imported module downloads only the languages,
themes, and modes actually used — matching how the worker build already
behaves. @pierre/diffs is split the same way so its pure patch parser (used by
the eager tool renderer) no longer drags the Shiki-importing render stack into
the startup graph.
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
* perf(ui): restore lazy heavy views and stop eager settings-graph loading
- MainLayout: DiffView/FilesView/GitView/PlanView return to lazyWithChunkRecovery
(they were silently made static in 2031e3b4 while their Suspense wrappers
remained), keeping the CodeMirror and @pierre/diffs stacks out of startup.
- ContextPanel: same lazy treatment for its Diff/Files/Git/Plan/Walkthrough
tabs, with null Suspense fallbacks.
- CommandPalette imported getSettingsNavIcon from SettingsView, statically
pulling the entire settings surface (SkillsPage -> CodeMirrorEditor -> vim
mode, theme registry -> @pierre/diffs) into the eager graph; the helper now
lives in lib/settings/metadata.
- The windowed SettingsWindow mounts only after its first open: rendering the
lazy component closed made React fetch the SettingsView chunk graph at
startup.
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
* perf(ui): keep @pierre/diffs + Shiki out of the eager chat graph and defer diff worker warmup
- DiffWorkerProvider no longer statically imports @pierre/diffs/worker or the
theme registry, and no longer spawns 3 workers plus a main-thread shared
highlighter during mount. Pools are created on demand through a dynamic
module load, warmed via requestIdleCallback after startup settles, and
useWorkerPool notifies consumers when a pool becomes available.
- ToolPart's rich diff preview moves to lazily loaded ToolPartDiffPreview;
the plain-text patch (PlainDiffFallback) renders while the chunk loads,
mirroring the existing error fallback. Theme registration happens during
render inside the lazy module so PatchDiff never renders unregistered ids.
- ChatInput mounts its lazy ToolOutputDialog only after the first attachment
preview opens instead of fetching the dialog chunk on the draft screen.
- getMarkdownSyntaxVars moves to a pierre-free markdownSyntaxVars module so
eager code-rendering consumers stop importing the registration module.
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
* perf(web): load ghostty-web and Nerd Fonts on first terminal use
- ghostty-web (638 KB raw JS + WASM VT) is dynamically imported when a
terminal actually mounts; TerminalView stays eagerly importable for the
bottom dock.
- The ~2 MB of CDN Nerd Fonts are no longer preloaded and force-loaded on
every cold start. index.html exposes an idempotent
__openchamberEnsureNerdFonts hook; TerminalViewport requests it on mount
and waits up to 2s so a cached font is in place before the glyph atlas is
built, while a cold CDN fetch never blocks the terminal. Runtimes without
the hook (VS Code, mobile) resolve immediately, matching their existing
fallback-font behavior.
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
---------
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
Worktrees are created with `git worktree add --no-checkout` and populated
with `git reset --hard`, neither of which runs git's post-checkout hook.
Invoke the hook explicitly after population with git's standard arguments
(null ref previous HEAD, checked-out HEAD, flag 1) and the worktree as cwd,
mirroring `git worktree add` without --no-checkout.
A missing or non-executable hook is skipped (matching git) and a failing
hook is logged as a warning, never failing worktree creation or session
bootstrap. Applied to both the web server git service and the VS Code
runtime's git service.
Show the running instance's local server URL and tunnel URL (when a
tunnel is active) as labeled, click-to-open buttons on the About page.
/api/system/info now reports the instance port and tunnel URL, resolved
lazily from the tunnel runtime so each Git-worktree instance identifies
itself in the UI without parsing terminal output.
Refs OPE-194
adm-zip <0.6.0 allows a crafted ZIP to trigger a ~4GB memory
allocation (GHSA-xcpc-8h2w-3j85). Bump the dependency in the web
and vscode packages to ^0.6.0. The new AdmZip(buffer) and
extractAllTo(dir, overwrite) APIs are unchanged, so no call-site
adaptation is needed.
Add a vitest regression test for the ClawdHub install path that
builds a real ZIP with adm-zip and asserts extractAllTo restores
files (including nested subdirectories) into the target skill dir.
Review follow-up:
- runtime.test.js: add syncProject wiring tests with a real temp-dir
project and real project-config runtime — asserts reconcileLoopTasks is
driven with the discovered loops when the project path is known (task
created, nextRunAt computed) and that plain listing is used when the
path cannot be resolved (reconcile not called).
- service.js: DELETE on a loop-owned task is rejected with a 400 only
while its loop file still exists on disk; once the file is gone the
orphan task can be deleted directly instead of waiting for the next
reconcile. Tests use real temp files for both branches.
- DOCUMENTATION.md: delete semantics updated accordingly.
- PR description refreshed for the final HEAD (test counts, reconciliation
contract, evidence wording).
Review follow-up:
- Reject loop files whose frontmatter name exceeds MAX_TASK_NAME_LENGTH
(80): task names are clamped at storage time, so a raw name longer than
the limit could never match the stored task identity. The file is treated
as malformed (definition: null) instead of creating an unreachable
definition; MAX_TASK_NAME_LENGTH is now exported from project-config.js
and shared with loops.js.
- Surface loop-sourced tasks in the scheduled-tasks dialog: tasks carrying
loopFile show a 'Managed by loop file <path>' note, and the enable
toggle / edit / delete actions are disabled with an explanatory tooltip,
since the file remains authoritative and would revert any such change.
run-now stays available. New locale keys added to all 11 message files
(i18n parity test enforces exact key sets).
- ScheduledTask type gains an optional loopFile field (additive, unknown
to older clients).
Review fixes for the markdown loop feature:
- Loop-owned tasks now adopt by loop file path, not task name, so renaming
a loop (frontmatter name or UI rename) renames the task in place instead
of leaving a stale duplicate that keeps running the old definition;
orphan duplicates of the same file are unscheduled.
- Unparseable loop files are reported to the scheduler as
definition:null entries: a task whose file still exists is kept with its
last good definition, and only a genuinely removed file unschedules it.
Transiently malformed files (mid-edit, bad merge) no longer delete tasks
or their runtime state.
- Adoption preserves UI-only execution fields (goalEnabled, goalTokenBudget,
permissionAutoAccept, variant) that the portable format does not define.
- DELETE on a loop-sourced task now returns 400 with guidance to remove the
loop file, instead of being silently undone by the next reconcile.
- Loops default to enabled: false; discovery of repository content never
auto-executes scheduled sessions unless the file explicitly enables them.
Regression tests for each fix; DOCUMENTATION.md updated.
Closesopenchamber/openchamber#2627
Listing a directory through a workspace symlink was returning realpath
entry paths. The file tree then rejected nested expand toggles because
those paths fall outside the workspace root.
Co-authored-by: Serhii Dziupin <makeittech@users.noreply.github.com>
Adds markdown-based scheduled-task definitions ("loops") discovered from
.agents/loops/*.md (project scope, ancestor directories up to the
worktree root) and ~/.agents/loops/*.md (user scope), mirroring the
skills discovery pattern.
File format: YAML frontmatter (name, schedule cron, enabled, model as
provider/model, optional agent/timezone) plus the markdown body as the
execution prompt. Discovery and parsing live in
scheduled-tasks/loops.js; project-config gains reconcileLoopTasks which
runs inside the project write lock on every syncProject:
- identity by task name; a loop takes over a matching task, preserving
its id and runtime state (markdown wins on conflict with JSON)
- tasks whose loopFile is gone are unscheduled; JSON tasks are never
removed
- new loops are created under deterministic loop:<scope>:<name> ids
- project scope shadows user scope on name collisions
- malformed files are skipped with a warning and never block valid ones
Runtime state stays in the project config/state store; it is never
written to the markdown files. Module documentation updated with the
file format and reconciliation rules.
Fixes#2583
When the managed OpenCode process exits but a server survives on the old
port (Windows: killProcessOnPort is a no-op, so the orphaned process tree
keeps the port), restartOpenCode() times out waiting for the port and
spawns a fresh server on a NEW port. HTTP/proxy traffic follows the new
port, but the global message-stream hub's upstream SSE reader stays pinned
to the old server's /global/event stream — that connection never closes —
so new events never reach the UI and the chat stops updating until the
app is restarted (#2638).
Lifecycle now fires an optional onOpenCodeRestarted hook after a
successful managed restart; index.js wires it to the new
messageStreamRuntime.rebindUpstream(), which restarts the shared hub
(its reader re-dials buildOpenCodeUrl → the current port) and closes
directory-scoped sockets so their per-connection readers rebuild against
the new port. External servers are untouched (their port cannot change).
Fixes#2638
OpenChamber's Settings → Chat → Small Model override only fed OpenChamber's
own /api/small-model/generate utility service; it never reached the managed
OpenCode server, whose internal title/summary generation reads small_model
from its config. With the override injected into OPENCODE_CONFIG_CONTENT at
managed-process launch, session title generation uses the user's explicit
model instead of falling back (or failing to resolve) — fixing sessions that
stayed untitled even with a Small Model configured.
Only an explicit override (smallModelUseDefault === false with a non-empty
smallModelOverride) is injected; "use default" leaves the config untouched
so OpenCode's own resolution chain stays authoritative. Malformed user config
is left unmodified. External OpenCode servers are unaffected (they are not
launched with this env).
Fixes#2497
parseMdFile now matches gray-matter (used by OpenCode) for file shapes
OpenChamber previously failed to parse: frontmatter whose closing '---'
sits at end-of-file without a trailing newline, a UTF-8 BOM prefix, and
YAML with unquoted colons in scalar values (via the same sanitizer
OpenCode applies). OpenCode parses these files, so OpenChamber must
too: otherwise the whole file was treated as the prompt body and a
save rewrote the existing YAML block into the body, prepending a
duplicate frontmatter block.
Refs OPE-178
The env var was already read and passed to the managed OpenCode server
spawn, but any non-empty string was accepted. Reject values that are
not a valid IP (IPv4/IPv6, brackets allowed) or DNS-style hostname with
a clear [config] error and fall back to the secure loopback default so
a typo can never silently bind a non-loopback address.
Refs OPE-231
The grand tunnel restructuring removed the CLI's auto-generated UI
password, so `openchamber -d --ui-password` (no value) silently started
an unauthenticated server instead of creating a password as in 1.8.1.
Restore generation for an explicit --ui-password flag without a value:
the password is generated before either launch path, passed to the
daemon/foreground process via OPENCHAMBER_UI_PASSWORD, persisted in the
instance state file, and surfaced once in human/quiet/json output.
Refs OPE-216
Follow-up to #2629, which stopped the walkthrough from comparing against a
branch that does not exist. The same guessing, and the same near-misses in how
the answer was applied, were left elsewhere:
- The default branch travelled as `rootBranchHint`, whose documented meaning is
"the branch the project root worktree is on". It gets its own option, because
a parameter that means two things is one the next caller gets wrong.
- A candidate equal to the branch being compared is skipped. In a plain checkout
the root hint *is* the current branch, so it won every time and produced a
comparison with itself; the repository default now wins there.
- The Changes and pull-request surfaces read the default branch too. A pull
request opened against a branch that does not exist is a worse failure than a
walkthrough that will not generate.
- `hasResolvableBaseBranch` matched `origin/feature/main` for a base of `main`,
passing the check and then failing the comparison it exists to prevent.
- `getRangeDiff` promoted only `origin/<base>`. A base carried by any other
remote stayed a bare name, which git resolves against refs/heads and nowhere
else, so it failed exactly as before.
- `getBranches` dropped every branch of a remote that did not answer, turning
"we could not ask" into "these branches are gone" — offline, that silently
removed comparisons that work fine against local remote-tracking refs.
- A remote with no `remote/HEAD` is asked once with `ls-remote --symref` rather
than falling back to the guess this data exists to replace.
The `defaultBranches` contract was documented under the status response; it
belongs to the branches response, which now has a section of its own.
Sérgio Pedro