The streaming status line and the finished turn footer are the same visual
line, but the swap used to jump: different font/color, a 2px left inset, a
20px row vs the footer's 32px (its h-8 action buttons set the height), and
the bottom-anchored chat pulling the line up because the finished message
carries more structure below its footer than the status row had.
Match the footer exactly: text-sm at muted-foreground/60, no left inset,
h-8 row, and mb-6 reserving the missing space below. Verified against the
live DOM: the footer appears at the exact pixel position the status row
occupied.
Neither failure came from the change that surfaced them; both suites were
asserting behavior the product had already moved.
issue-2039 mocks `session-actions`, and session creation's authoritative
steps live there: the created session becomes current under the directory
the server confirmed, and that directory enters sync's routing index. The
mock did neither, so every assertion about current session and directory
routing read an empty store. It now performs both, which is what makes
those assertions test the store's real resolution order again.
issue-2903 pinned `<ChatEmptyState />` inside ChatContainer's empty-and-idle
branch. The draft transition animation moved that surface to the draft, and
the branch now returns a neutral placeholder. What the test is actually for
— the idle branch not rendering a status row of its own — is kept.
A project could pin the model new chats start on, but not the level to
run it at: the default cascade dropped any variant as soon as a project
model won, and only ever considered the global one — which belongs to
the global model.
Projects now carry `defaultVariant` alongside `defaultModel`, stored and
sanitized only next to that model, and the cascade passes it through.
Both controls sit in one "Defaults for new chats" group laid out like the
Sessions defaults, and the level appears only for models that offer them.
Changing the default model, variant or agent wrote the chat's live
selection unconditionally. For a session carrying its own choice the
effect was mostly transient — ModelControls restores that choice on the
next hydration — but a message sent right after the change went out on
the new model.
The guard is the same pair of signals ModelControls restores from
(shouldPreserveManualModelOverride): a manual selection plus a saved
selection for the current session. Chats following the default still
switch immediately.
Picking a project in Settings called setActiveProject, which relocates
the chat, the session list, the file tree and the Git surface. Reading
another project's MCP servers or agents moved the user's whole app.
It had to, because the configuration stores resolved the directory
themselves from the active project and held one flat list. Each of them
now takes an explicit directory — omitted still means the active project,
so every caller outside Settings is unchanged — and keys loaded data by
directory next to a flat mirror of the active project. Chat, autocompletes
and pickers keep reading that mirror; a load for another directory writes
only the map. A failed load restores that directory's previous list.
Settings resolves its own directory through useSettingsDirectory, backed
by a session-local settingsProjectPath that follows the active project
until the user picks something else.
The visibility hook looked the context panel up under the directory the
card reports about. A managed Chat reports about none, so the lookup ran
against an empty key and answered "closed" while the panel was open.
It now resolves the key the way the rail and the panel do, through
useEffectiveDirectory. The reporting directory is no longer an input to
this hook at all.
The selected provider on the Providers settings page shared a field with
the chat's model resolution. Changing the chat model or agent, and
`loadAgents` reconciling a selection, all wrote to it, so the page jumped
to an unrelated provider while the user was reading or editing one.
Only the sidebar click writes it now. A refresh no longer re-derives it
either: a provider missing from a refreshed list is expected while
OpenCode restarts and plugin-registered providers re-register.
Adding an SSH connection now starts from the hosts in the SSH config
instead of a blank command field, ports and install options sit behind
Advanced settings, and each connection reports one of three states with
the failure text and an action that resolves it.
Remote installs no longer touch the root-owned global npm prefix: npm is
pinned to a prefix under $HOME and bun is resolved at its known location,
because an SSH login shell exposes neither on PATH. The opencode CLI is
resolved the same way and handed to the remote server through
OPENCODE_BINARY, and the server is started and stopped through the
resolved binary rather than PATH — the HTTP shutdown route sits behind UI
authentication and never stopped anything.
A managed remote server can also be published to the remote machine's own
network. That requires a UI password, enforced in the form and again in
the SSH manager.
Plain Enter now inserts a newline in the expanded composer
Cmd/Ctrl+Enter still sends the message
Helps avoid accidental sends while writing long prompts
Truncation was measured once on mount and watched through a ResizeObserver
on the clamped element, whose own height stops changing under line-clamp.
A message measured before its content settled never regained the expand
affordance, and a just-sent one measured before its turn finished laying
out could not be expanded at all.
Measure at click time so expanding depends on whether the text is clipped
right now, re-read once on the next frame after mount, and observe the
children (which keep their natural height) plus subtree swaps.
Preserve layout with invisible placeholders while worker highlighting loads
Reuse cached highlighted lines on first render when available
Differentiate loading from failed highlighting results for better fallback behavior
The additional context and pull request template blocks ran into the file
list without a blank line, and the template's markdown headings blended into
the surrounding instructions. Each block now starts on its own, and template
content is fenced with explicit markers labelled as structure to reuse rather
than instructions to follow.
Claude-Session: https://claude.ai/code/session_01NzK8hD1hFHHHwv2FgZJwC2
The server echo of an optimistic user message carries different part
ids, so the reducer dropped the optimistic part and appended the server
one at the end, and MessageBody keyed user parts by part id. The key
change remounted the text subtree (blank frame, markdown re-parse,
truncation state reset) and the append reordered text against file
parts.
Replace the optimistic part in place and key user parts positionally.
Commit generation now samples the last 10 commit subjects and instructs the
model to match their language, prefix convention, and length, falling back to
Conventional Commits only when the history shows no consistent style.
PR generation probes the conventional pull request template locations and
reuses the repository's own sections and checklists as the body structure when
one is found.
Claude-Session: https://claude.ai/code/session_01NzK8hD1hFHHHwv2FgZJwC2
Prevent managed Chat drafts and sessions from inheriting an active project's repository context in the Work Status panel while retaining directory-independent status sections.
Parakeet is an offline model trained on whole utterances, so re-decoding
the growing buffer to animate a live transcript cost O(n^2) work for a
result the final decode replaced. Sessions now decode once per committed
segment, and the composer shows a scrolling waveform of the mic level
instead of running text.
Long dictations split at a pause once past 60s (hard cap 90s) instead of
on a blind 15s timer, so cuts no longer land mid-word. Committed segments
decode while the user is still speaking: a 185s dictation returns 4.1s
after stop instead of 11.0s, with identical text (816 vs 817 words).
Also fixes two ways the stream manager could silently drop transcribed
audio. It now counts the commits it issued instead of trusting the
session's echoed events, so a commit still in flight when the client
finishes can no longer be left out of the final text. And segment
byte/peak accounting is reset where the commit is issued rather than when
the event arrives, which could mistake the tail of a dictation for
silence and clear it.
Remove retired Command Code, Discord, and Telegram integration entries, search targets, and documentation. Keep Command Code provider usage and logo support available through normalized provider ID aliases.
Show every change on the current branch relative to its base in the
Changed/Staged/Last turn dropdown. The base comes from the branch's
reflog record or an explicit per-branch user choice (persisted), never
a main/master guess; when git has no record the user picks a base once
from a searchable branch list.
- server: GET /api/git/branch-base (reflog-derived base),
GET /api/git/range-files (name-status -z with rename/copy
destination paths and -C copy detection)
- shared UI: optional getBranchBase/getGitRangeFiles runtime APIs
with boundary parsing; persisted per-branch overrides keyed by
runtime+directory+branch
- DiffView: branch scope with confirmed-unavailability coercion of
persisted tabs (detached HEAD, default-branch checkout, metadata
settled without a default), range-invalidated diff cache guarded
against stale completions, bounded branch-metadata retry, read-only
diff actions in branch scope; hidden in VS Code
- helper module branchDiffScope.ts with tests for coercion,
availability, race conditions, and retry exhaustion
Opens the chat tab when a new inline comment is submitted
Automatically focuses the chat composer for quicker follow-up
Leaves draft updates unchanged when editing an existing comment
Render managed Chats through the shared folder-aware session group and keep folder ownership at the Chats root, including new drafts and older per-directory scopes.
Hide worktree actions for Chats, make worktree shortcuts inert on Chat drafts, and let new-session shortcuts inherit an active session directory while explicit Chats controls still create managed Chat drafts.
Track Chats and Recent section sentinels so the desktop sticky overlay changes identity only when the corresponding section reaches the top of the sidebar.
Create projectless chat sessions under a managed, date-scoped Chats directory and clean abandoned or deleted session folders.
Add Chats to sidebar state, startup cache, shared context, and Electron Mini Chat while keeping VS Code project-only. Resolve managed chat directories to one server-side memory owner and document the runtime contracts.
Adds a mandatory early load step for the communication-style skill.
Removes the blanket always-on trigger from the skills table.
Clarifies that the guidance applies to all task output.
Adds a new communication-style skill for rewriting text in a more human voice
Replaces the old unslop skill with the new name and guidance
Updates agent instructions to apply the communication-style skill to every task
* feat(skills): remove ClawHub catalog integration
Drop the ClawHub registry as a skills catalog source across web server,
shared UI, VS Code, docs, and locales. The catalog now serves git-based
sources only: the curated Anthropic repo and user-defined repositories.
Also removes the now-unused adm-zip dependency.
* feat(skills): redesign catalog around curated GitHub repositories
Replace the single-source dropdown with a card grid of curated GitHub
repositories (Anthropic, OpenAI, Cursor pstack/skills, Matt Pocock) plus
user-defined sources. Source cards show skill counts, GitHub stars, and
last-updated time; a global search covers all loaded sources.
Server: curated sources gain GitHub repo metadata (stars, pushed_at)
fetched best-effort with a 3-hour in-memory and on-disk cache; scans
run through a concurrency-limited, deduplicated cache with 3-hour TTL
persisted across restarts. Refresh still bypasses the cache.
Shared UI: source cards, global search with clear button, per-skill
GitHub links, install/installed states. VS Code curated list updated
to match. All new copy translated across 12 locales.
* fix(skills): address catalog review findings
- GitHub metadata fetch timeout drops to 1.5s (under the catalog
client's 3s deadline) and failed lookups cache briefly (5 min) so
repeated catalog loads do not re-hit a failing API.
- Disk cache files are written with owner-only permissions (0o600);
rename preserves the mode.
- loadSource deduplicates concurrent in-flight requests per source and
the shared isLoadingSource flag now clears only when the last active
source load finishes.
Defines a process for rewriting text to sound more human
Lists patterns to remove, including AI-style phrasing and overused punctuation
Adds practical guidance for clearer, more direct prose
Plugin providers are registered from a plugin's `config` hook and credentialed
from its `auth` loader, both inside the running OpenCode process. Nothing about
them reaches `opencode.json` or `auth.json`, so resolution that only reads files
could not see them: selecting such a model failed with "has no known API base
URL" while the same model worked in chat (#2666).
`GET /provider` is where that state is visible. A new `runtime-providers`
module keeps one cached snapshot of it and reports, per provider, the
credential and endpoint OpenCode itself resolved. Credential resolution becomes
config -> runtime -> auth.json, and endpoint resolution config -> openai default
-> runtime -> models.dev catalog.
Providers with a dedicated wire format (Copilot, ChatGPT-plan OpenAI, Anthropic,
Google) are excluded from the runtime credential: for them OpenCode reports an
OAuth access token that their real transport does not accept.
opencode zen is excluded when the user has no zen login. OpenCode then reports
the sentinel `apiKey: "public"` and trims its catalog to free models that run on
its own infrastructure; the sentinel is never read as a credential.
Claude Code stays refused for background actions even when a plugin publishes an
OpenAI-compatible endpoint for it, because that endpoint is a facade over the
Claude Agent SDK and spawns the CLI per request.
No capability probe. Asking `GET /models` does identify a plugin whose protocol
lives in its own `fetch`, but measured across the 166 providers with an `api`
URL in the models.dev catalog it also denies six that work and simply have no
`/models` route. A provider that vanishes from the picker explains nothing,
while one that fails on use says why, so availability stops at credential and
endpoint.
The same list drives the Small Model and Changes Walkthrough pickers.
Validated against a real OpenCode with four plugin providers loaded: offered
providers went from 3 to 7, zen and Claude Code stayed out, and a generation
through a plugin-backed model that previously failed now returns.
Shows the auth panel only when provider credentials are actually required
Keeps custom providers defined in config from being treated as incomplete
Adds coverage for the new provider auth check
Adds new integrations docs for Claude Code, Command Code, and Cursor
Links the Providers pages to the new integrations guide
Updates the docs sidebar with localized integrations navigation
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.
A failed nightly run left edits in the maintenance clone, and every later run
correctly refused to work on a dirty worktree, so one failure stalled the whole
pipeline until morning.
Maintenance task commands now recognise a gitignored .maintenance-clone marker.
In a marked disposable clone they discard leftover debris, return to main, and
continue; in a human working copy they still stop and touch nothing.
Add an explicit abort protocol: revert your own edits, confirm the worktree is
clean, release the claim, and report. Restore the honest skip that the
complete-file rule had squeezed out, since a laundered fix is worse than a
documented skip, and describe how to handle a file that is entirely an
external-data boundary instead of inventing generic JSON contracts.
Maintenance task commands now fill .github/PULL_REQUEST_TEMPLATE.md section by
section instead of inventing their own headings, and follow-up tasks keep the
description true for the final HEAD while preserving hand-added content.
Raise the anti-slop batch window to 60-120 findings and require each selected
file to be finished: remaining findings need an individual specific reason,
shared root causes count once, and difficulty alone no longer justifies a skip.
A half-fixed file otherwise returns as a second pull request over the same code.
Add the maintenance-review command, which reviews every open anti-slop and
react-doctor pull request and fixes the findings directly rather than
commenting, without merging or approving.
Vendor the anti-slop Oxlint plugin at tools/oxlint/anti-slop and register it
in oxlint.config.ts, with Oxlint's own rule categories disabled so ESLint
stays the general-purpose linter.
Add scripts/anti-slop.mjs (bun run deslop) mirroring the React Doctor batch
interface: next-batch, check-batch, active, release, top, file. Batch handoff
directories now double as file claims shared across clones via
~/.openchamber/maintenance-claims, so concurrent maintenance batches from
either pipeline never select the same file.
Harden both scheduled maintenance flows: stop on a dirty worktree, stop on
NO BATCH AVAILABLE, validate per package instead of workspace-wide, and pin
react-doctor to 0.9.12. The anti-slop task command documents concrete
good and bad fixes and forbids laundering types to satisfy a rule.
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.
Bohdan Triapitsyn