Upstream changes detected: ACP spec, Gemini CLI, GitHub Copilot CLI, Qwen Code

[github-actions]

This issue was opened automatically by upstream-watch.yml on 2026-05-18 because at least one upstream project this plugin depends on shipped a new release.

Detected changes

ACP spec — v0.12.2 → v0.13.2

• Release: v0.13.2 (published 2026-05-17)
• Why this matters for the plugin: Defines the JSON-RPC methods the plugin's acp-client.mjs implements. Spec changes can introduce new methods we should handle (e.g. terminal/*, session/request_permission, cursor/ask_question).

Release notes excerpt

Fixed

• (unstable) Update additionalDirectories guidance (#1227)

Gemini CLI — v0.41.2 → v0.42.0

• Release: Release v0.42.0 (published 2026-05-12)
• Why this matters for the plugin: Drives plugins/multi/scripts/lib/adapters/gemini.mjs. Watch for changes to ACP handshake, model alias resolution, MCP support, or new approval modes.

Release notes excerpt

What's Changed

• fix(cli): prevent automatic updates from switching to less stable channels by @Adib234 in fix(cli): prevent automatic updates from switching to less stable channels google-gemini/gemini-cli#26132
• chore(release): bump version to 0.42.0-nightly.20260428.g59b2dea0e by @gemini-cli-robot in chore(release): bump version to 0.42.0-nightly.20260428.g59b2dea0e google-gemini/gemini-cli#26142
• fix(cli): pass node arguments via NODE_OPTIONS during relaunch to support SEA by @cocosheng-g in fix(cli): pass node arguments via NODE_OPTIONS during relaunch to support SEA google-gemini/gemini-cli#26130
• fix(cli): handle DECKPAM keypad Enter sequences in terminal by @Gitanaskhan26 in fix(cli): handle DECKPAM keypad Enter sequences in terminal google-gemini/gemini-cli#26092
• docs(cli): point plan-mode session retention to actual /settings labels by @ifitisit in docs(cli): point plan-mode session retention to actual /settings labels google-gemini/gemini-cli#25978
• fix(core): add missing oauth fields support in subagent parsing by @abhipatel12 in fix(core): add missing oauth fields support in subagent parsing google-gemini/gemini-cli#26141
• fix(core): disconnect extension-backed MCP clients in stopExtension by @cocosheng-g in fix(core): disconnect extension-backed MCP clients in stopExtension google-gemini/gemini-cli#26136
• Update documentation workflows with workspace trust by @g-samroberts in Update documentation workflows with workspace trust google-gemini/gemini-cli#26150
• refactor(acp): modularize monolithic acpClient into specialized files by @sripasg in refactor(acp): modularize monolithic acpClient into specialized files google-gemini/gemini-cli#26143
• test: fix failures due to antigravity environment leakage by @adamfweidman in test: fix failures due to antigravity environment leakage google-gemini/gemini-cli#26162
• fix(core): add explicit empty log guard in A2A pushMessage by @adamfweidman in fix(core): add explicit empty log guard in A2A pushMessage google-gemini/gemini-cli#26198
• feat(cli): add --delete flag to /exit command for session deletion by @AbdulTawabJuly in feat(cli): add --delete flag to /exit command for session deletion google-gemini/gemini-cli#19332
• test(core): add regression test for issue for ToolConfirmationResponse by @Adib234 in test(core): add regression test for issue for ToolConfirmationResponse google-gemini/gemini-cli#26194
• Add the ability to @ mention the gemini robot. by @gundermanc in Add the ability to @ mention the gemini robot. google-gemini/gemini-cli#26207
• test(evals): add EvalMetadata JSDoc annotations to older tests by @akh64bit in test(evals): add EvalMetadata JSDoc annotations to older tests google-gemini/gemini-cli#26147
• fix(core): reduce default API timeout to 60s and enable retries for undici timeouts by @Adib234 in fix(core): reduce default API timeout to 60s and enable retries for undici timeouts google-gemini/gemini-cli#26191
• fix(core): distinguish fallback chains and fix maxAttempts for auto vs explicit model selection by @adamfweidman in fix(core): distinguish fallback chains and fix maxAttempts for auto vs explicit model selection google-gemini/gemini-cli#26163
• fix(cli): handle InvalidStream event gracefully without throwing by @adamfweidman in fix(cli): handle InvalidStream event gracefully without throwing google-gemini/gemini-cli#26218
• ci(github-actions): switch to github app token and fix bot self-trigger by @gundermanc in ci(github-actions): switch to github app token and fix bot self-trigger google-gemini/gemini-cli#26223
• Respect logPrompts flag for logging sensitive fields by @lp-peg in Respect logPrompts flag for logging sensitive fields google-gemini/gemini-cli#26153
• fix: correct API key validation logic in handleApiKeySubmit by @martin-hsu-test in fix: correct API key validation logic in handleApiKeySubmit google-gemini/gemini-cli#25453
• fix(agent): prevent exit_plan_mode from being called via shell by @Abhijit-2592 in fix(agent): prevent exit_plan_mode from being called via shell google-gemini/gemini-cli#26230

• Fix: Inconsistent Case-Sensitivity in GrepTool by @gemini-cli[bot] in # Fix: Inconsistent Case-Sensitivity in GrepTool google-gemini/gemini-cli#26235

• docs(core): add automated gemma setup guide by @Samee24 in docs(core): add automated gemma setup guide google-gemini/gemini-cli#26233
• Allow non-https proxy urls to support container environments by @stevemk14ebr in Allow non-https proxy urls to support container environments google-gemini/gemini-cli#26234
• fix(bot): productivity and backlog optimizations by @gundermanc in fix(bot): productivity and backlog optimizations google-gemini/gemini-cli#26236
• refactor(acp): delegate prompt turn processing logic to GeminiClient by @sripasg in refactor(acp): delegate prompt turn processing logic to GeminiClient google-gemini/gemini-cli#26222
• fix(cli): refine platform-specific undo/redo and smart bubbling for WSL by @cocosheng-g in fix(cli): refine platform-specific undo/redo and smart bubbling for WSL google-gemini/gemini-cli#26202
• fix: suppress duplicate extension warnings during startup by @cocosheng-g in fix: suppress duplicate extension warnings during startup google-gemini/gemini-cli#26208

…(release notes truncated; click the link above for the full text)

GitHub Copilot CLI — v1.0.44 → v1.0.48

• Release: 1.0.48 (published 2026-05-14)
• Why this matters for the plugin: Drives plugins/multi/scripts/lib/adapters/copilot.mjs. Watch for ACP changes, slash-command additions/removals, or auth flow changes.

Release notes excerpt

2026-05-14

• Model picker displays actual token prices instead of dot indicators for token-based billing users
• Instruction files with unquoted glob patterns in applyTo frontmatter (e.g. applyTo: */.ts) are now applied correctly
• Input text with CJK characters or emoji renders without blank gaps between lines
• /context shows correct token limits for all models instead of always showing 128k
• Auto-disable the built-in github-mcp-server in Azure DevOps-only workspaces when running in prompt/headless mode, matching interactive mode behavior
• Terminal cursor positions correctly on the input field instead of on decorative elements like the selected tab
• ACP clients receive updated config options when the active model is changed
• /ask dialog no longer prompts for follow-up replies it cannot receive
• Skill content injected to the model no longer includes YAML frontmatter metadata

Qwen Code — v0.15.10 → v0.15.11

• Release: Release v0.15.11 (published 2026-05-13)
• Why this matters for the plugin: Drives plugins/multi/scripts/lib/adapters/qwen.mjs. Watch for ACP support changes (the --acp flag graduated from --experimental-acp recently).

Release notes excerpt

What's Changed

• chore(release): v0.15.10 by @qwen-code-ci-bot in chore(release): v0.15.10 QwenLM/qwen-code#4019
• perf(core): bound session-list metadata reads to head/tail 64KB; pool buffer; lazy message count by @qqqys in perf(core): bound session-list metadata reads to head/tail 64KB; pool buffer; lazy message count QwenLM/qwen-code#3897
• test: stabilize main e2e flakes by @yiliang114 in test: stabilize main e2e flakes QwenLM/qwen-code#3992
• feat(cli): core built-in i18n coverage by @shenyankm in feat(cli): core built-in i18n coverage QwenLM/qwen-code#3871
• feat(core): write runtime.json sidecar for active sessions by @yeelam-gordon in feat(core): write runtime.json sidecar for active sessions QwenLM/qwen-code#3714
• feat(telemetry): inject traceId/spanId into debug log files for OTel correlation by @doudouOUC in feat(telemetry): inject traceId/spanId into debug log files for OTel correlation QwenLM/qwen-code#3847
• feat(tools): defer low-frequency built-in tools to reduce initial prompt size by @wenshao in feat(tools): defer low-frequency built-in tools to reduce initial prompt size QwenLM/qwen-code#4022
• ci: skip unnecessary release and SDK checks by @yiliang114 in ci: skip unnecessary release and SDK checks QwenLM/qwen-code#3984
• feat(installer): add standalone archive installation by @yiliang114 in feat(installer): add standalone archive installation QwenLM/qwen-code#3776
• feat(cli): Ctrl+B promote keybind (#3831 PR-3 of 3) by @wenshao in feat(cli): Ctrl+B promote keybind (#3831 PR-3 of 3) QwenLM/qwen-code#3969
• feat(cli): add --json-schema for structured output in headless mode by @wenshao in feat(cli): add --json-schema for structured output in headless mode QwenLM/qwen-code#3598
• refactor(cli): remove legacy qwen auth CLI subcommand, redirect to /auth TUI dialog by @pomelo-nwu in refactor(cli): remove legacy qwen auth CLI subcommand, redirect to /auth TUI dialog QwenLM/qwen-code#3959
• chore(deps): upgrade ink 6.2.3 → 7.0.2 + bump Node engine to 22 by @chiga0 in chore(deps): upgrade ink 6.2.3 → 7.0.2 + bump Node engine to 22 QwenLM/qwen-code#3860
• feat(skills): Add codegraph skill for PR review risk analysis and conflict detection by @BingqingLyu in feat(skills): Add codegraph skill for PR review risk analysis and conflict detection QwenLM/qwen-code#3910
• refactor(core): route side-query LLM calls through runSideQuery chokepoint by @tanzhenxin in refactor(core): route side-query LLM calls through runSideQuery chokepoint QwenLM/qwen-code#3775
• feat(tools): keep ask_user_question always-visible to surface clarification UX by @tanzhenxin in feat(tools): keep ask_user_question always-visible to surface clarification UX QwenLM/qwen-code#4041
• feat(core): improve Anthropic proxy compatibility and enable global prompt cache scope by @wenshao in feat(core): improve Anthropic proxy compatibility and enable global prompt cache scope QwenLM/qwen-code#4020
• fix(cli): keep long model stats header on one line by @Jerry2003826 in fix(cli): keep long model stats header on one line QwenLM/qwen-code#4032
• refactor(telemetry): remove dead useCollector setting and unreachable TelemetryTarget.QWEN by @doudouOUC in refactor(telemetry): remove dead useCollector setting and unreachable TelemetryTarget.QWEN QwenLM/qwen-code#4061
• chore(core): runtime.json sidecar follow-ups from #3714 review by @wenshao in chore(core): runtime.json sidecar follow-ups from #3714 review QwenLM/qwen-code#4030
• fix(test): repair stale --json-schema integration assertion by @tanzhenxin in fix(test): repair stale --json-schema integration assertion QwenLM/qwen-code#4075
• feat(cli): add tools.toolSearch.enabled setting for prefix-caching models by @wenshao in feat(cli): add tools.toolSearch.enabled setting for prefix-caching models QwenLM/qwen-code#4069
• feat(core): replace fdir crawler with git ls-files + ripgrep fallback by @scrollDynasty in feat(core): replace fdir crawler with git ls-files + ripgrep fallback QwenLM/qwen-code#3214
• fix(cli): improve rendering on narrow terminals by @chiga0 in fix(cli): improve rendering on narrow terminals QwenLM/qwen-code#3968
• feat(dashscope): support DASHSCOPE_PROXY_BASE_URL for prompt cache via API gateway by @HeZiGang in feat(dashscope): support DASHSCOPE_PROXY_BASE_URL for prompt cache via API gateway QwenLM/qwen-code#3991
• Upgrade GitHub Actions for Node 24 compatibility by @salmanmkc in Upgrade GitHub Actions for Node 24 compatibility QwenLM/qwen-code#1876
• doc[sdk-python] Expand Python SDK usage documentation by @doudouOUC in doc[sdk-python] Expand Python SDK usage documentation QwenLM/qwen-code#3995
• ci(e2e): stabilize MCP/CLI flows and cancel stale main runs by @yiliang114 in ci(e2e): stabilize MCP/CLI flows and cancel stale main runs QwenLM/qwen-code#4039
• fix(channels): expand tilde in channel cwd config by @qqqys in fix(channels): expand tilde in channel cwd config QwenLM/qwen-code#4045

…(release notes truncated; click the link above for the full text)

Also worth checking (manual — Copilot, please research these too)

• Cursor agent CLI — changelog: https://cursor.com/changelog · forum (bug reports): https://forum.cursor.com/c/bug-report/6
  • Drives plugins/multi/scripts/lib/adapters/cursor.mjs. The plugin currently works around the 2026.04.17-787b533 ACP regression (see maybeWarnAboutCursorVersion and ensureCursorAllowlist). When Cursor ships a fix, both workarounds can likely be simplified or removed.

Plugin files most likely to need updates

• plugins/multi/scripts/lib/acp-client.mjs (shared ACP JSON-RPC client; buildAutoApproveRequestHandler)
• plugins/multi/scripts/lib/acp-terminals.mjs (client-side terminal services)
• plugins/multi/scripts/lib/mcp-servers.mjs (MCP wiring for ACP session/new)
• plugins/multi/scripts/lib/adapters/{codex,gemini,cursor,copilot,qwen}.mjs (per-CLI adapters)
• plugins/multi/scripts/multi-cli-companion.mjs (companion runtime + dispatch)

What I'd like you to do, @copilot

1. Read the linked release notes for each detected change above, plus the manual-reference changelogs.
2. Compare what changed against the relevant adapter / shared code in this repo. Look specifically for:
   • New ACP methods we should handle in acp-client.mjs's buildAutoApproveRequestHandler
   • Renamed / deprecated CLI flags or model IDs hardcoded in any adapter
   • New CLI capabilities that obsolete a workaround we currently ship (e.g., the Cursor 2026.04.17 regression workaround in cursor.mjs — check if a newer Cursor release fixes it, and if so, propose removing maybeWarnAboutCursorVersion and the allowlist-injection or scoping it tighter)
   • Breaking changes that would silently break the plugin
3. For each change that warrants action, open a focused PR against master with the minimal fix. Reference the upstream release / commit / forum thread in the PR description.
4. If a detected change does NOT need any plugin update, reply on this issue with a short note saying "no plugin updates needed for X — reason: …" and close it.
5. If something is ambiguous (you can't tell from release notes whether the plugin is affected), ask in a comment rather than guessing.

You may use ACP_TRACE=1 and the rest of the diagnostic patterns documented in plugins/multi/skills/customize/SKILL.md if you want to verify behavior empirically.

---

State tracked in .github/upstream-state.json — bumped by this same workflow. If you want to suppress a noisy upstream from this watch, edit .github/scripts/upstream-watch.mjs.