The panel stored notes, todos and plans inside one shared JSON file that
six unrelated domains also wrote to, synchronised itself through window
CustomEvents, and could only read plans. It is now Project knowledge:
server-owned storage with explicit routes, a store with rollback, a
section sidebar, plans that open and edit in place, and search across
all of it.
Notes and plans the user pins travel with every message sent in that
project. Pinning is project state, not an attachment to one message, so
it holds until unpinned and the work status panel names what is riding
along and can detach it.
Agent memory is added alongside, in two scopes: what is true about the
user, and what is true about this codebase. The split is not cosmetic —
a wrong project fact costs one project and is noticed, while a wrong
global fact quietly shapes every session everywhere and the user has no
code to check it against. It stays separate from notes so an agent
mistake cannot land in what the user wrote. Sessions receive an index of
titles only; bodies are read on demand, because an index carrying full
text grows until it crowds out the conversation.
Deciding what a session must be told, and whether it has been told, now
lives on the server. The client owned it before, which meant sessions
started without a UI — scheduled tasks, sessions the agent dispatches —
received nothing at all, and a tab's record of what it had sent outlived
the conversation: after compaction the agent no longer held the block
while the tab went on believing it did. What was delivered is recorded
in the session's own metadata, and compaction restores it through the
runtime that already restores pinned messages, in the same turn.
Agent memory ships dark behind OPENCHAMBER_MEMORY_ENABLE: unset, there
is no tool, no routes, no session index, no settings row and no panel
tab. Absent rather than switched off, so nothing invites turning on a
feature that has not been announced. Pinned notes and plans are
unaffected and ship as normal.
Sending a message without a provider/model selection failed silently:
only a console.warn was emitted and the UI gave no feedback. Surface a
toast (i18n key chat.chatInput.toast.noModelSelected, added to all 11
locales) so the user knows why the message was not sent.
Review follow-up on the #2769 highlight caches.
Fingerprint strength. The block/highlight caches are now global and
content-addressed, so a hash collision no longer mis-colors a block — it returns
a *different* block's rendered HTML and shows the user source they never wrote.
Length + one 32-bit FNV-1a is not enough key space for that failure mode at
session scale. `contentFingerprint` now combines two independent 32-bit
multiplicative hashes with a final avalanche (~64 bits); two multiplies per
character are free next to Shiki tokenization.
Streaming churn. Content addressing made every streaming step of the trailing
`live` block insert a new cache entry, so one long message evicted the settled
`full` blocks the fix exists to keep warm. `full` and `live` blocks now use
separate caches; the live cache is small (32 entries / 2MB) because it only has
to absorb repeat renders of the same step.
Redundant worker-side caches. `markdown-worker.ts` is the only sender to the
Shiki worker, and its client cache is larger than the worker-side ones, so the
worker caches could not serve a hit the client had not already served — they
only duplicated up to 48MB of payloads in a second heap. Removed; the reason
memoization belongs on the client is now documented there, along with why only
`highlightTokens` carries a theme in its key.
Dead `cacheKey` plumbing. `renderMarkdownBlocks` kept a `cacheKey` parameter it
only `void`-ed. Removed it and the now-unused `useMorphdomMarkdown` prop; the
remaining call-site local is renamed `fadeKey` for what it actually keys.
Tests: image-mode cache identity, streaming-does-not-evict-settled-blocks,
live-cache reuse, and a 20k same-length-source fingerprint collision check.
Each new guard was verified to fail without its fix.
Conflict: packages/ui/src/components/chat/markdown/markdownCore.ts
main added per-image-mode markdown parsers (`imageMode` threaded through
`parseBlock` and into the block cache key); this branch replaced the
identity-keyed block cache with a content-addressed LRU. Resolution keeps the
content-addressed cache and folds `imageMode` into the content key, so the
`inline` and `label` renderings of the same source cannot answer for each other.
* fix(chat): defer composer value writeback during IME composition
The controlled-writeback effect compared the value prop against the
CodeMirror document and, on mismatch, dispatched a wholesale replacement
with the caret forced to the end. While the browser composes (pinyin,
kana, hangul) the uncommitted text lives in the DOM, not in the document,
so the mismatch is expected and the dispatch interrupted the IME session
and jumped the cursor. Skip the writeback while the view is composing,
using CodeMirror's public compositionStarted getter; the composition
commits through its own pipeline and reports via onChange.
Fixes#2527
* fix(chat): preserve external composer writes during IME
* fix(chat): restore composition-wide writeback guard
---------
Co-authored-by: Bohdan Triapitsyn <artmore@protonmail.com>
Adds a per-project forced provider (github|gitlab|gitea) on top of the
per-project API base URL overrides: stored under gitProviders.provider in
projects/<projectId>.json, sanitized server-side, and winning over remote-host
detection both in useGitProvider and in server repo resolution
(parseGitLabRemoteUrl/parseGiteaRemoteUrl accept any host when the provider
is forced). The Projects settings page replaces the three always-visible URL
fields with a provider selector (auto-detect + the three forges) and one URL
override for the active provider. Global provider override fields on the
GitHub/GitLab/Gitea settings tabs now render only once an account is
connected, and Settings search availability matches that gating.
Also fixes the useConfigStore/useDirectoryStore circular-import TDZ in the
bundled chunk via the window-registered store handle and defers the directory
subscription to a microtask; fixes the Gitea PR merge payload (Do carries the
merge-style string enum, not a boolean + MergeMethod); and adds a documented
Gitea client live-test harness (scripts/gitea-live-test.ts + client.d.ts).
Per provider (github|gitlab|gitea) a project override stored in
~/.config/openchamber/projects/<projectId>.json under gitProviders wins
over the global settings.json value (precedence: project override > global
> built-in default). Server forge routes resolve the override per request
directory (worktree-aware via git-common-dir + containment + path fallback,
60s TTL cache); the override host is also accepted for remote parsing and
client detection. New GET/PUT /api/projects/:projectId/git-providers route;
client openchamberConfig preserves the server-owned gitProviders key;
Projects page gains a Git provider API base URLs section; detection store
hydrates per-project overrides (memory-only, server-authoritative).
Configure a default API base URL per git provider (github/gitlab/gitea)
in settings.json gitProviders, with GitHub Enterprise support (Octokit
baseUrl + device-flow web origin derived from the API base), and replace
the client-side custom-domains list with server-persisted detection URL
chips (SSH/HTTPS forms normalized to hosts). The configured API base host
auto-counts as a detection host. Settings round-trip through the existing
/api/config/settings sanitizer; the UI store hydrates from server settings
with a one-time localStorage migration.
The SDK client fetch wrapper now applies a 30s timeout to non-streaming
reads. Without it, a socket that neither resolves nor rejects keeps the
directory bootstrap concurrency slot busy forever and the UI stays on
"loading sessions". Long-lived streams (POST prompts, the /event SSE)
are explicitly excluded so they are not cut off mid-flight.
The normalized "request timed out" error is added to the retry
allowlist alongside undici's "terminated" (the exact failure observed
in #2470 when undici tears down a half-open upstream connection); both
are transient while the managed OpenCode process restarts. Caller-
initiated aborts keep their original error shape so a user-cancelled
request is not retried.
Tests cover: GET timeout fires after the bound, POST is not timed out,
/event SSE is not timed out, caller abort wins, AbortError is not
retried, and the SDK normalized error is retried 3x.
The assignee/label/milestone combobox and @-mention list were inline
absolute-positioned inside the ContextPanel rail (overflow-hidden) and the
PR-view scroll container, so the panels were clipped and invisible. Render
both via createPortal to document.body with fixed positioning computed from
the trigger's viewport rect (viewport-edge clamping, flip to the other side
when there is no room), close on outside pointerdown/scroll/resize while
still allowing interaction inside the panel, and keep the panel inside the
outside-click check.
Also show the loading spinner until the first lookup completes instead of a
brief 'No matches' flash (useForgeLookup now reports 'initialized').
The Gitea create-issue path (server route, client, wire api, facade, UI
button/dialog) landed with the forge user-lookup work; close the remaining
verification/documentation gaps:
- wire: gitea.test.ts covers issueCreate (POST /api/gitea/issues/create,
full/optional body, error throw)
- docs: DOCUMENTATION.md lists createIssue client method and the
POST /repos/{owner}/{repo}/issues endpoint (labels as names)
- ui: fix stale 'read-only by design' JSDoc in GiteaIssuesSection — creation
is offered here and the detail view provides edit/close/reopen
Repo-scoped assignable-user search for GitHub, GitLab, and Gitea, surfaced as
an assignee combobox in the metadata editor and @-mention autocomplete in
forge comment/reply/review surfaces.
- server: GET /api/{provider}/users/search (assignees / project members),
query + directory/override repo resolution, 429 -> 503, connected:false
degradation; GitLab assignee writes resolve login -> ID server-side
- wire: searchUsers (+ searchLabels/milestones/branches/tags) on the three
API clients with tests
- facade: userSearch capability (all three), searchUsers adapters,
mapGithubAssignee/mapGitlabMember/mapGiteaAssignee -> ForgeUser
- ui: ForgeLookupCombobox (keyboard nav, debounced 30s-TTL cache,
connected-only caching), ForgeMentionTextarea (@ token parsing, caret
restore), free-text fallback when lookup is unavailable; i18n in 12 locales
- extras sharing the same infrastructure: GitLab create-issue dialog and
label/milestone/branch/tag lookups in the metadata editor
- linkedIssues: provider/repo/host fields (backward-compatible), gitea id generation fix, parseLinkedIssueRef/parseForgeEntityUrl
- work-status panel: live state cards (TTL-cached), link-by-URL dialog, per-row unlink
- forge entity view: 'Chats working on this' section derived from the session store
Add the Gitea teapot mark (Simple Icons, 24x24) to the icon sprite as a
custom icon, and swap the gitea fallback icons for it wherever the
provider brand is shown: context rail PR surface, context panel PR tab,
GitHeader PR chip, and the work-status Gitea PR row.
Extract a single shared parseGitHost (packages/ui/src/lib/gitHost.ts) used
by both the custom-domain registry (normalizeProviderDomain) and remote
detection (gitProvider). Previously both parsers required a user@ prefix,
rejecting valid bare scp remotes like codeberg.org:owner/repo.git, and
mangled IPv6 hosts. Now scp forms with or without a user, ssh:// URLs with
ports, git+ssh scheme, and bracketed/unbracketed IPv6 all normalize to a
bare hostname; Windows-path-like input is rejected.
Add an Issues tab alongside the merge-request / pull-request content in the
context panel's PR surface for both providers. The tab lists open issues
lazily and expands a selected issue into an inline detail view (body, labels,
assignees, comments) with a back action. Read-only: no issue create, update,
or close actions. Adds 15 i18n keys across all 11 locales.
The token breakdown of an assistant message accumulates across every API
round-trip inside the turn: each tool call re-reads the whole cached
prompt, so input/cache.read add up to several times the context window.
Every context-usage surface summed those fields, which is why the meter
could read 330% of a 1M window whose real fill was 232,872 tokens
(23.3%), and why reopening an older session jumps the readout (#2562).
The server reports the final round-trip's window as tokens.total
(optional in the message schema; opencode 1.18.18 returns it, verified
against its live /session/:id/message API). Prefer it everywhere the
window fill is displayed and fall back to summing only when the server
did not send it: contextTokensFromBreakdown in tokenUtils now owns that
rule, and the context store extractor, sync store getter, work status
panel, context sidebar, VS Code layout, mini chat, and mobile metadata
all use it instead of their own inline sums.
Fixes#2562
Branch status resolves an open PR across the whole fork network first, so a
merged fork PR can never hide an open upstream PR for the same head. Only when
no target has an open PR does the branch's newest closed/merged PR come back,
as history.
The panel shows that history as a compact note and offers creating the next PR
below it, instead of either sticking on a terminal PR or going blank after a
merge. Terminal associations stay persisted for reload continuity but are never
treated as authority: they revalidate on the discovery cadence and on focus.
History is looked up only for the branch's own remote and name, and remembered
per repo+branch, so the extra lookup cannot exhaust the route's resolve budget.
The checks summary and merge-permission lookup are skipped for a closed or
merged PR, where neither is actionable.
Auto-derived project labels were title-cased, turning .ssh into .Ssh and
opencode-claude into Opencode Claude. Show the folder name verbatim in the
sidebar, window title, settings selector and notification templates, and
migrate persisted legacy labels back to the folder name (manual renames
are preserved).
Create no longer aborts when recoverStaleDraftDirectory rewrites the
implicit new-chat draft to the active project during the create probe.
Also rank the Unreleased Chat bullet below the changelog highlights.
Co-authored-by: serkraser <serkraser@gmail.com>
Keep both Unreleased changelog bullets: the lastDirectory new-chat
fallback and the Git PR panel stale-status fix from main.
Co-authored-by: serkraser <serkraser@gmail.com>
Bohdan Triapitsyn
bot-hermes