Commit Graph

16443 Commits

Author SHA1 Message Date
Peter Steinberger dabeae8c90 perf(test): tighten Teams lifecycle polling (#109686) 2026-07-16 22:52:25 -07:00
Arseniy Palagin 5e13f459a7 fix(telegram): preserve forward origins in debounce batches (#108985)
* fix(telegram): preserve forward origins in debounce batches

* fix(telegram): deliver batched forward origins to agent

* refactor(telegram): keep debounce forward cleanup local

Co-authored-by: Arseniy Palagin <263096044+aspalagin@users.noreply.github.com>

---------

Co-authored-by: Arseniy Palagin <263096044+aspalagin@users.noreply.github.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 22:47:44 -07:00
Peter Steinberger 630b16805c fix(providers): refresh zai/kimi/moonshot/xai catalogs against July 2026 vendor docs (#109666)
* fix(providers): refresh vendor catalogs and setup hints

* test(providers): align registration metadata expectations
2026-07-16 22:41:03 -07:00
Peter Steinberger a1dfd47edd refactor(sdk): collapse plan updates to typed steps (#109660)
* refactor(sdk): collapse plan updates to typed steps

Per owner decision, remove the shipped steps: string[] SDK field and its unreleased planSteps replacement early. Collapse onPlanUpdate to one typed steps field. Retain wire-level string normalization for external Codex plugin version skew.

* chore(sdk): tighten rebased surface budget
2026-07-16 22:35:29 -07:00
Peter Steinberger dfe9e55452 perf(test): remove Google Chat timeout proof sleep (#109671) 2026-07-16 22:31:24 -07:00
Alix-007 6bd15d00c4 fix(openai): prevent device-code login hangs on stalled requests (#109494)
* fix(openai): bound device code requests

* fix(openai): preserve proxy routing for device code auth

* test(openai): simplify device proxy assertions

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 22:26:56 -07:00
YangManBOBO a7aa85293e fix(qqbot): stop misrouting mixed-case group/channel targets to the C2C API (#109008)
* fix(qqbot): accept mixed-case type tags in target parser

* refactor(qqbot): centralize target type parsing

Co-authored-by: 潘晓波0668000512 <pan.xiaobo@xydigit.com>

* style(qqbot): format target parser

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 22:25:03 -07:00
marchpure 6b4f372ff6 fix(codex): expose assistant completion idle timeout config (#97233)
* fix(codex): expose assistant completion idle timeout config

* chore: rerun codex idle timeout config checks

* docs(codex): document assistant completion idle timeout

* test(codex): cover assistant idle timeout default

* fix(codex): preserve runtime option compatibility

---------

Co-authored-by: marchpure <marchpure@users.noreply.github.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 22:22:38 -07:00
Yuval Dinodia f6582dfb5e fix(file-transfer): denyPaths does not deny listing the denied directory itself (#109233)
* fix(file-transfer): deny listing the directory a denyPaths rule names

A denyPaths entry like "**/.ssh/**" did not deny dir.list on the denied
directory itself, so listing /home/me/.ssh returned its full contents,
including filenames such as id_rsa and authorized_keys. Only paths strictly
under the directory were denied.

The deny gate matched the requested path against the globs as a plain string,
and a trailing /** requires something after the separator, so the bare
directory never matched. Every command routes its path through
evaluateFilePolicy, but dir.list is where it is observable, since its payload
is the directory's own contents. dir.fetch was incidentally safe because it
rechecks each returned entry, and those entries do match; dir.list has no
such recheck.

Match the deny globs against the directory form of the path at that one gate.
The check stays deny-side only, so it can fail closed but never open.

* test(file-transfer): cover denied directory path forms

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 22:22:01 -07:00
Peter Steinberger a5237fe925 feat(llama-cpp): gate Gemma default by RAM (#109585) 2026-07-16 22:17:57 -07:00
Peter Steinberger 5199bfafea fix(line): preserve webhook events through dispatch failures (#109655)
* test(line): remove obsolete replay-cache coverage

* fix(line): durably spool webhook events

Co-authored-by: NianJiuZst <180004567+NianJiuZst@users.noreply.github.com>

* test(line): tighten webhook spool lifecycle proof

* fix(line): use canonical turn adoption lifecycle

Co-authored-by: NianJiuZst <180004567+NianJiuZst@users.noreply.github.com>

* refactor(line): keep spool types private

Co-authored-by: NianJiuZst <180004567+NianJiuZst@users.noreply.github.com>

---------

Co-authored-by: NianJiuZst <180004567+NianJiuZst@users.noreply.github.com>
2026-07-16 22:13:26 -07:00
Wynne668 f63ea72f28 fix(github-copilot): prevent failed usage checks from holding connections (#109540)
* fix(github-copilot): release failed usage responses

* test(github-copilot): preserve usage HTTP errors

Co-authored-by: ZengWen-DT <ceng.wen@xydigit.com>

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 22:09:53 -07:00
wahaha1223 0e96e85b19 fix(signal): release connections after failed CLI downloads (#109442)
* fix(signal): cancel failed CLI download responses

* fix(signal): close all failed installer responses

Co-authored-by: wahaha1223 <0668001153@xydigit.com>

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 22:01:53 -07:00
Peter Steinberger e873a7f955 refactor(memory): move QMD coordination to SQLite (#109636)
* refactor(memory): move QMD coordination to SQLite

* chore: keep release notes in PR body

* chore: annotate lease SQLite primitive
2026-07-16 22:01:34 -07:00
Peter Steinberger 7be5d78fd0 feat(workspaces): add full-bleed single-widget tabs (#109627)
* feat(workspaces): add full-bleed app tabs

* fix(workspaces): harden full-bleed rendering

* fix(workspaces): eliminate full-bleed iframe gap

* fix(workspaces): keep tab layout type internal

---------

Co-authored-by: Eva <eva@100yen.org>
2026-07-16 21:56:57 -07:00
wahaha1223 0c25b52f7c fix(huggingface): release connections after failed model discovery (#109464)
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:54:17 -07:00
Monkey-wusky b98b8e7313 fix(googlechat): cancel non-OK cert fetch response body before throwing (#109519)
* fix(googlechat): cancel non-OK cert fetch response body before throwing

When fetchChatCerts receives a non-OK response from the Google Chat
cert endpoint, it throws without cancelling the response body.
Cancel the body before throwing, matching the pattern established
in reef guard adapters (#109196) and fetchOpenRouterModels.

* test(googlechat): cover cert response cancellation

Co-authored-by: 毛宇豪0668001457 <mao.yuhao@xydigit.com>

* test(googlechat): satisfy strict call-order types

Co-authored-by: 毛宇豪0668001457 <mao.yuhao@xydigit.com>

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:52:45 -07:00
Peter Steinberger a77c0770ff refactor(speech-core): share TTS request pre-resolution across discord and voice-call (#109640) 2026-07-16 21:47:06 -07:00
Ayaan Zaidi 19b5e564d7 fix(channels): clear type-aware lint debt and re-measure the drain seam after the split 2026-07-17 10:12:54 +05:30
Ayaan Zaidi 84c8b23835 refactor(channels): drop ingress exports orphaned by the whatsapp split 2026-07-17 10:12:54 +05:30
Ayaan Zaidi 08db31f7fa test(telegram): relocate non-fence dispatch coverage
Lifecycle invariants move to core contract tests; relocated dispatch suites
keep draft finalization, status reactions, and pipeline-init cleanup coverage
without fence machinery, and polling/webhook suites assert core-drain
semantics (#108656).
2026-07-17 10:12:54 +05:30
Ayaan Zaidi 1126ffba7f test(telegram): prove drain-backed transport invariants
Transport proof for the migrated ingress path: offset-after-spool-ack,
webhook redelivery on write failure, update_id/lane mapping, and the
authorization-gated supersede suite including paired-DM, owner, topic-override
and other-bot-command cases (#108656).
2026-07-17 10:12:54 +05:30
Ayaan Zaidi f7786a16cf refactor(telegram): drive ingress through the core drain
Both transports now enqueue durably and pump the shared drain: polling keeps
offset-after-committed-enqueue, webhook keeps 200-after-spool-write. Deletes
the reply fence, private claim-owner, retry policy, and per-transport
claim/replay/watchdog loops. Dispatch outcomes propagate: failed-retryable
releases for replay instead of tombstoning. Supersede policy stays
Telegram-owned, authorization-gated with pairing-store and owner identities
via the same resolver as normal ingress; room_event ambient pending remains
supersedable by a later user turn (#108656).
2026-07-17 10:12:54 +05:30
Ayaan Zaidi 16c14e5bbf feat(channels): expose the turn-adoption lifecycle seam
Replace the bare onTurnAdopted callback and queuedFollowupLifecycle with one
turnAdoptionLifecycle surface (onSettled guaranteed via finally; adoption-loss
aborts queued steering turns without transcript replay), threaded through the
turn kernel and agent runner, exposed to plugins via
runtime.state.openChannelIngressDrain and the channel-outbound SDK barrel,
with surface budgets re-measured against the narrowed baselines (#108656).
2026-07-17 10:12:54 +05:30
badgerbees fedcffc4ae fix(memory-lancedb): stop forwarding embedding dimensions upstream (#69707)
* fix(memory-lancedb): truncate embeddings locally

* fix(memory-lancedb): tighten dimensions validation in config schema and manifest

* fix(memory-lancedb): require integer embedding dimensions

* fix(memory-lancedb): retry embeddings without dimensions on rejection

* fix(memory-lancedb): narrow dimensions fallback

* fix(memory-lancedb): recognize tuple-style provider rejection for dimensions

* refactor(memory-lancedb): tighten dimensions fallback

* style(memory-lancedb): format dimensions fallback

* fix(memory-lancedb): preserve embedding timeout budget

* test(memory-lancedb): type vector search mock

* fix(memory-lancedb): reject invalid dimension values

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:35:44 -07:00
Peter Steinberger e3618454f2 refactor(talk): add realtime voice session harness and adopt in google-meet and voice-call (#109497)
* refactor(talk): add realtime voice session harness and adopt in google-meet and voice-call

* fix(google-meet): satisfy prefer-const in realtime engine wiring
2026-07-16 21:28:44 -07:00
Wynne668 7b6c6a0802 fix(acpx): preserve Unicode in catalog tool arguments (#109589)
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:21:04 -07:00
Peter Steinberger d275942136 test: accelerate active-memory timeout cleanup coverage (#109631) 2026-07-16 21:19:11 -07:00
LZY3538 233922b5d8 fix(anthropic): accept Claude CLI apiKeyHelper auth (#97492)
* fix(anthropic): accept Claude CLI apiKeyHelper auth

Co-authored-by: Lzy10033538 <liu.zhenye@xydigit.com>

* fix(anthropic): make synthetic auth fallback explicit

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:16:41 -07:00
Wynne668 5f56a541d3 fix(opencode): preserve Unicode in catalog tool input (#109591)
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:10:18 -07:00
snowzlmbot 17c2ce05d8 fix(secrets): keep startup alive when TTS SecretRefs are missing (#101265)
* fix(secrets): degrade missing TTS SecretRefs at startup

* test(secrets): keep non-activating startup strict

* test(secrets): mark denied key fixture synthetic

* test(secrets): use synthetic TTS key fixture

* test(secrets): use neutral TTS key placeholder

* test(secrets): isolate TTS key placeholder

* test(secrets): shorten TTS ref fixture name

* test(secrets): normalize synthetic credential fixtures

* test(secrets): isolate optional redaction coverage

* fix(secrets): preserve degraded TTS ref ownership

* refactor(secrets): keep optional resolver internal

* test(secrets): cover default provider alias misses

* test(secrets): pin explicit provider ownership

* style(secrets): format optional assignment imports

* refactor(secrets): keep optional metadata private

* style(secrets): restore collector file header

* fix(secrets): isolate unavailable SecretRef owners

Co-authored-by: snowzlmbot <293528334+snowzlmbot@users.noreply.github.com>

* test(secrets): complete provider fixtures

* style(status): avoid degraded path shadowing

* style(secrets): satisfy runtime lint

* refactor(secrets): keep error codes internal

* fix(secrets): keep unowned assignments fail closed

* fix(secrets): preserve provider resolution batching

* fix(secrets): normalize stalled resolution errors

* fix(secrets): reject provider limit violations

---------

Co-authored-by: snowzlmbot <293528334+snowzlmbot@users.noreply.github.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:07:01 -07:00
Mert Başar 9d6133a931 fix(agents): block fallback completion delegation (#92011)
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 21:06:40 -07:00
Peter Steinberger 96815a437b fix(channels): centralize access gates and media fallback (#109548)
* refactor(channels): shared supplemental sender gating, allowlist-match adoption, outbound mop-ups

* fix(plugin-sdk): skip-aware media sequence with text fallback for empty URLs

* chore(plugin-sdk): align surface budgets after rebase

* test(qqbot): type media sender mock calls

* fix(plugin-sdk): distinguish empty media sequences

* fix(plugin-sdk): track void media sends
2026-07-16 20:55:08 -07:00
Peter Steinberger b1541c933f test: accelerate Codex startup timeout coverage (#109614) 2026-07-16 20:43:21 -07:00
EVA 09575bd20f feat(workspaces): show agent status and custom-widget decisions (#101826)
* feat(workspaces): add operations widgets

* fix(workspaces): format agent progress translation

* fix(workspaces): keep operations widget helpers private

* feat(workspaces): harden operations widgets

Co-authored-by: EVA <eva@100yen.org>

* chore(i18n): defer generated locale refresh

* refactor(workspaces): isolate approval state

* fix(release): declare direct checklist execution

* fix(release): satisfy checklist declaration lint

* fix(release): align checklist declaration order

---------

Co-authored-by: Eva <eva@100yen.org>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 20:19:49 -07:00
Yuval Dinodia a06799ab2a fix(acpx): keep leaked non-openai model out of the Codex ACP thinking slot (#95852)
* fix(acpx): keep leaked non-openai model out of the Codex ACP thinking slot

Codex ACP spawn mis-routed an inherited non-OpenAI fleet default into the
reasoning-effort slot and aborted (#95780). Replace the splitter with a closed
classifier and make the spawn path provenance-aware: drop an inherited leaked
default so Codex starts on its own default, but fail closed with
ACP_INVALID_RUNTIME_OPTION when a caller explicitly selects an unsupported or
malformed model. Thread a modelExplicit flag from resolveAcpSpawnRuntimeOptions
through the ACP runtime ensure contract; strip it before the acpx delegate.

Dropping the inherited default only at ensureSession was not enough: the manager
still persisted the leaked model in runtimeOptions, and the first turn replayed
it through applyRuntimeControls -> setConfigOption(model), which the new
fail-closed Codex control path then rejected. ensureSession now reports the
effective model it applied on the returned handle (applied | dropped), and the
manager persists that effective model, so a dropped inherited default is never
saved or replayed as a model control before the first turn. Explicit unsupported
selections still fail closed at spawn and never persist.

* test(acp): split runtime config validation tests

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 20:15:03 -07:00
clawsweeper[bot] 1c5c3d5201 fix(codex): before_prompt_build receives conversation messages (#101752)
Co-authored-by: clawsweeper <274271284+clawsweeper[bot]@users.noreply.github.com>
Co-authored-by: xingzhou <275915537+zhangguiping-xydt@users.noreply.github.com>
2026-07-16 20:10:25 -07:00
Peter Steinberger 28a3540f32 feat: add native inline widget support (#109212)
* feat: add native inline widget support

* refactor: simplify plugin surface refresh

* fix: preserve plugin surface refresh API

* fix: reload widgets after WebKit termination

* fix: harden native widget refresh

* fix: bound native widget recovery

* docs: defer native widget release note

* chore: refresh native i18n inventory

* fix: harden native widget delivery

* refactor: simplify native widget API surface

* fix: serialize native widget capability refresh

* fix: harden native widget recovery

* fix: recover native widget capabilities

* style: format widget refresh role selection

* fix: correct widget refresh formatting

* fix: harden cross-platform widget recovery

* fix: bind widget trust and recovery to routes

* chore: refresh native i18n inventory

* chore: regenerate Android gateway protocol

* test: remove unused release workflow read
2026-07-16 19:24:41 -07:00
wings1029 730cfd774d fix(qa): bound Windows taskkill process termination (#109453)
`taskkill.exe` can hang when the target process is stuck in a kernel wait
state (e.g. a pending I/O request). Add a 5-second timeout so QA scenario
cleanup fails closed instead of hanging the test runner indefinitely.

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 19:11:43 -07:00
wings1029 13e43d4c05 fix(qa): bound gateway child taskkill termination (#109454)
`taskkill.exe` can hang when the target process is stuck in a kernel wait
state. Add a 5-second timeout to both taskkill invocations in the gateway
child process tree cleanup so stalled termination fails closed instead of
hanging the QA runner.

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 19:10:53 -07:00
Peter Steinberger 658b601ee5 feat(llama-cpp): in-process local GGUF text inference provider (#109444)
* feat(llama-cpp): add in-process text inference

* test(llama-cpp): narrow setup provider fixture

* fix(llama-cpp): trim public surface and refresh docs map

* fix(llama-cpp): import Context type in inference test
2026-07-16 18:53:55 -07:00
xingzhou afd0fe32c2 fix(qa-lab): preserve boundary emoji in gateway startup diagnostics (#104840)
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 18:26:26 -07:00
EVA 368941ae7a feat(workspaces): add secure preview widget (#101354)
* feat(workspaces): add secure preview widget

* fix(workspaces): type preview reload assertion

* fix(workspaces): harden preview widget

* fix(workspaces): honor preview URL binding ownership

* chore(workspaces): remove unused widget exports

---------

Co-authored-by: Eva <eva@100yen.org>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 18:21:54 -07:00
Peter Steinberger 0a19e1ed58 refactor(channels): centralize mention and group activation decisions (#109480)
* refactor(channels): adopt shared mention-gate decision; consolidate group activation

* fix(matrix): carry effective mention decision

* test(session): pin qqbot accessor migration
2026-07-16 18:15:58 -07:00
Wynne668 54f1800d2a fix(matrix): bound recovery key stdin input (#108120)
Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 18:14:33 -07:00
Peter Steinberger 784ede0af1 feat(system-agent): local-model viability — context cap, thinking off, route-aware timeout (#109445)
* feat: improve system agent local model viability

* fix: forward Ollama effective context cap

* fix(system-agent): keep flat 120s agent-turn budget

* fix(system-agent): keep manifests timeout helper module-local
2026-07-16 18:11:22 -07:00
Peter Steinberger 92146f9f80 refactor(talk): share audio-energy stats and speech-threshold gate across voice surfaces (#109466) 2026-07-16 18:10:26 -07:00
Peter Steinberger edd052989b refactor(google-meet): merge host and node realtime engines behind one audio transport (#109413)
* refactor(google-meet): merge host and node realtime engines behind one audio transport

* refactor(google-meet): internalize engine helpers left test-only after realtime-node removal
2026-07-16 17:39:53 -07:00
Peter Steinberger d71c1fe596 feat(onboarding): prefer strongest local model in guided detection (#109250)
* feat(onboarding): rank detected local models

* docs(sdk): comment local-model rank bucket layout

* feat(sdk): rank gemma4 first for local setup-assistant models
2026-07-16 17:30:49 -07:00
sunlit-deng a8dc0683cc fix(xai): ignore blank audio env api key (#108953)
* fix(xai): ignore blank audio env api key

* test(xai): focus blank env regression

Co-authored-by: sunlit-deng <yang.jiajun1@xydigit.com>

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>
2026-07-16 17:24:35 -07:00