Three defects proven by running the shipped configs:
Kubernetes probes checked only the status code against /startupz, but the
pinned image predates that route and the Control UI answers unknown paths
with a catch-all 200. A wedged pod was therefore marked Ready forever.
Probes now assert the JSON probe contract and target routes the pinned
image actually serves; verified in a kind cluster where the old command
exits 0 on the missing route and the new one exits 1.
render.yaml set no dockerCommand, so the image CMD ran without
--allow-unconfigured and a fresh Render disk exited 78 with 'Missing
config' before binding. Reproduced locally with Render's exact env.
The Cloudflare Container readiness poll had the same route mismatch
against operator-supplied official image digests; it now polls /healthz,
which every published image serves.
Also replaces an R2 verification step that could never fail: wrangler
cannot list object keys, so the documented command 404'd into || true.
* fix(ui): defer GitHub project clone until submit
Keep remote project selection as draft state and materialize it before sessions.create on Start Session.
* fix(ui): consolidate deferred project selection
* test(ui): drop duplicate remote project assertion
* refactor(config): retire dead streaming.progress.render key
The key had zero runtime consumers after #122552. Core doctor now strips it via stripRetiredTuningKnobs, and production LOC is net -13.
* fix(tooling): pin plugin SDK surface counts to the reduced export set
The retired progress-draft render reader counted twice via channel-outbound and channel-message's wildcard re-export.
* fix(skills): keep ClawHub publisher identity from search through install
ClawHub search returns one entry per publisher, so several results can share a
slug. Every client collapsed the selection to that bare slug before calling
skills.detail and skills.install, and ClawHub answered 409 AMBIGUOUS_SKILL_SLUG
with no in-product way forward.
searchClawHubSkills now records the publisher-qualified reference once, on the
result that carries it, and the Gateway protocol documents it. skills.detail
parses the same reference grammar skills.install already accepted, so review and
install cannot resolve to different publishers. Control UI carries that one
reference through row actions, detail, busy state, and acknowledgement retries,
and shows it so otherwise identical rows are distinguishable.
Fixes#117633
* fix(apps): send the ClawHub publisher reference from native skill browsers
macOS, iOS, and Android read the qualified reference from search results and use
it for skills.detail, install, busy state, installed matching, and list identity,
so two publishers sharing a slug stay distinct instead of collapsing into one
ambiguous request.
* fix(skills): refuse external-source skill detail instead of reading a same-slug skill
ClawHub has no source-qualified read endpoint, so a skills-sh reference parsed
down to its bare slug would have returned a registry skill's card while install
resolved the external artifact. Review and install could name different skills.
skills.detail now fails closed on any reference that carries a source, and the
macOS and AgentPro rows show the publisher reference next to the summary instead
of only when a summary is missing, so same-slug rows stay distinguishable.
* chore(apps): refresh native i18n source baseline for the skill row references
* refactor(skills): drop the unread search-result ownerHandle field
installRef is the one reference clients send back, and no client reads the
publisher handle separately, so the protocol and Control UI carry one field
instead of two.
* fix(skills): name the next step when external skill detail is refused
Clients that gate install behind a successful review would otherwise see only a
refusal, so the error names the direct install path and the CLI equivalent.
* fix(macos): use a doc comment on the ClawHub row subtitle
swift-format's docComments rule requires doc comments on declarations; the
subtitle property carried a regular comment and failed macos-swift.
* fix(skills): carry ClawHub trust state to clients that can install
Forwarding installRef let clients install the exact publisher the operator
picked, including external skills-sh sources. It did not forward the trust
state that says ClawHub never scanned that source, so iOS AgentPro — the one
surface that installs in a single tap with no review step — could install an
unscanned artifact with nothing on screen saying so. The CLI already labels
these (docs/clawhub/cli.md, docs/cli/skills.md); native clients could not,
because trustState was never on the wire.
trustState becomes an optional field on SkillsSearchResultSchema. It is purely
additive: older clients ignore an unknown key and the field is absent for
registry results, so downgraded readers are unaffected and no protocol version
moves.
Every client that renders a search row now shows "Not scanned by ClawHub",
matching the CLI wording exactly: iOS AgentPro in the row above the install
button, macOS and Android beside the review action, and Control UI on the row
that explains why review is refused for these sources.
Covered by a wire assertion that the state reaches clients for an external
source and stays absent for registry rows, plus decode-and-label tests on the
shared Swift kit and the Android parser, and a Control UI render assertion.
* fix(ui): size the ClawHub detail dialog to a refusal message
Refusing detail for an external source made an error-only dialog reachable.
The shared preview panel reserves a tall reader height for skill documents, so
a two-line refusal rendered in a mostly empty dialog and read as broken rather
than deliberate. Found by inspecting the review captures.
* revert(ui,apps): drop the ClawHub trust label layer
Maintainer product decision: skills.sh runs its own scanners, so OpenClaw does
not add a second alert layer in the apps. Removes the label from Control UI,
iOS, macOS and Android, and drops the trustState wire field that nothing would
render. The CLI keeps its existing label; changing that is a separate call.
Publisher identity, the fail-closed detail refusal, and the message-only dialog
are unchanged. Splits the oversized skills view test file to satisfy max-lines
without a suppression.
* test(ui): fix ClawHub skill fixture checks
* chore(plugin-sdk): refresh API baseline
---------
Co-authored-by: Patrick Erichsen <patrick.a.erichsen@gmail.com>
* feat(ui): open subagent details in chat sidebar
* chore: drop changelog edit (release generation owns it)
* refactor(ui): drop duplicate close in subagent detail panel
The sidebar region header already owns a Close Details control in both
wide and narrow layouts; the panel-local X duplicated it 40px away.
* fix(ui): stop subagent transcript loader when pane presentation retires
Pane retention wipes sidebarContent directly, so the detail slot's
render-time reset can never run again; a pending refresh timer plus
incoming task events kept refetching chat.history for a hidden panel.
* docs(ui): note close-control ownership in subagent detail header
* fix(ui): break transcript renderer import cycle
* fix(ui): use shared action cursor for subagent rows
* fix(ui): keep composer model picker stable while typing
* test(ui): prove typing keeps model picker stable
Co-authored-by: Colin Johnson <colin@solvely.net>
---------
Co-authored-by: Peter Steinberger <steipete@gmail.com>
* fix(sessions): zero highWaterBytes no longer clears all session history
resolveHighWaterBytes passed an explicit non-positive highWaterBytes through
verbatim. The resolved value is the disk-budget cleanup loop's stop condition,
so a zero target made enforce mode evict every unprotected session and prune
its extracted archives instead of trimming to the documented 80% default.
Route the non-positive case to the function's existing unusable-value branch
(computeDefault). Not null: that disables the budget and permits unbounded
growth, which is right for a cap but wrong for a target.
Sibling of #119422, which fixed the same harm for maxDiskBytes and guarded
only resolveMaxDiskBytes.
* test(infra): isolate worktree migration discovery
Keep worktree migration coverage focused on its real filesystem, Git, and SQLite owner while avoiding unrelated channel and plugin doctor cold starts on fork CI.
Co-authored-by: masatohoshino <g515hoshino@gmail.com>
* test(ci): carry owner-approved SDK and doctor gate repairs
Carry the already-approved plugin SDK contract manifest and focused doctor-flow test isolation from the maintainer-owned CI repair. Preserve real config migration, persistence, snapshot, and SQLite cleanup coverage; no production behavior changes.
Co-authored-by: masatohoshino <g515hoshino@gmail.com>
* fix(sessions): use the renamed withTestDir helper in the new budget test
* fix(sessions): align high-water zero contract
* style(sessions): format high-water changes
---------
Co-authored-by: Peter Steinberger <steipete@gmail.com>
Co-authored-by: FullerStackDev <263060202+fuller-stack-dev@users.noreply.github.com>
* fix: prompt caching breaks on Claude Opus 5 and Sonnet 5
shouldPreserveThinkingBlocks() matched modern Claude ids with a hardcoded
family list plus /claude-[5-9]/ future-proofing. That regex assumes the
generation follows the prefix (claude-5-x), but shipped generation-5 ids
place the family in between (claude-opus-5, claude-sonnet-5, claude-mythos-5),
so none matched and dropThinkingBlocks flipped to true. Signed thinking blocks
were stripped from replayed history, diverging every request and invalidating
the Anthropic prompt cache on the default opus/sonnet aliases.
Read the generation from the id instead, preserving blocks for generation 4
and newer. This handles both id shapes and matches the idiom already used in
audit-extra.sync.ts and live-model-filter.ts.
Closes#121251
* fix: correct Claude thinking replay contract
Use the canonical Claude model identity and exact Anthropic preservation contract when deciding whether replay may retain prior thinking blocks. Carry deployment metadata through provider-family and fallback paths, and include canonicalModelId in the transcript policy cache key.
Release note: Restore prompt-cache reuse for Claude Opus 5 and Sonnet 5 while continuing to strip unsupported Sonnet and Haiku 4.5 thinking history.
* refactor: remove unused replay compatibility helper
Use the canonical Claude drop predicate at the remaining test call sites and remove the unconsumed preservation export so the dependency/dead-code gate stays clean.
* fix: keep Claude replay contract internal
Use canonical llm-core identity resolution without exporting a new llm-core capability, avoiding unintended Plugin SDK API drift.
---------
Co-authored-by: FullerStackDev <263060202+fuller-stack-dev@users.noreply.github.com>
The template shipped without the operator-facing half: no way to prove the
restore path works before depending on it, no cost signal for the always-on
versus webhook-only choice, no log surfaces, and no failure-mode index.
Adds an architecture diagram, a restore rehearsal, measured recovery numbers
from the real-R2 proof, provisioned-resource cost guidance, wrangler tail
observability, and a troubleshooting index covering image architecture,
Litestream S3 credentials, the startupz/readyz split, and ephemeral-disk
config loss.
* fix(agents): keep session model selection canonical
Preserve configured session model identity across response aliases and turn-local fallbacks.
* fix(gateway): align session model projection consumers
* test(gateway): mark session event model selection
* fix(gateway): cascade session stop to subagents
Cascade exact and session-wide aborts through the subagent registry and keep Control UI Stop visible while descendants remain active.
* refactor(gateway): absorb stop cascade orchestration
* chore(gateway): refresh plugin SDK API contract for abort refactor
Regenerates the export-closure contract after the stop-cascade refactor
moved gateway abort orchestration exports, and drops a redundant Boolean()
wrapper flagged by lint. Public SDK surface gate is unchanged.
* chore(gateway): regenerate plugin SDK API baseline
A steer rejected on the run/leaf fence parks as a failed steered queue row.
Retry previously refused whenever the target run was gone, leaving the
user's message permanently stuck with 'that run is no longer active'.
Retry now converts the parked row into a plain queued send when the session
is idle, so the text delivers as a fresh turn; active-run retries keep the
original steer binding, and disconnected retries keep the existing error.
Adds hard policy: UI-visible changes need before/after screenshots or a
short video as PR evidence; gateway-behavior changes provable in the
Control UI need an isolated dev-gateway live run with a recorded video.
Documents the verified video upload flow (user-attachments endpoint,
mp4/webm both served) and the bare-URL embed rule for GitHub's player.
* feat(ui): show environment facts in Where picker
Preserve environment platform, trust, session-host, and capability metadata through new-session discovery while keeping placement availability tied to live executable nodes. Render bounded quiet device/cloud facts and add mocked browser coverage for task #33.
* docs(plan): track picker environment facts
* refactor(ui): split place browser rendering
* test(microsoft-foundry): remove global test bridge
* test(microsoft-foundry): type boundary fixtures
* perf(ui): keep route transition out of startup
The PR-changed test planner fails safe to the compact full-suite plan for any diff touching packages/**, but that compact plan excludes all extension test configs, so mixed package+extension PRs landed with zero extension test execution (escapes: PR #120534 breaking extensions/codex run-attempt.native-hook-relay.test.ts, PRs #122163/#121522 and cd7b7f639d breaking media-understanding-provider.test.ts and thread-lifecycle.test.ts on main full runs). The preflight now appends whole-config shards for the diff's touched extensions whenever the precise plan fails safe; whole configs (not precise targets) because the fail-safe cause leaves the non-extension diff's extension impact unbounded.
Peter Steinberger