v1.2.0

What's Changed

• fix(ui): batch fix four UI bugs — AskUserQuestion, scroll arrow, workspace wizard, file preview by @davidliuk in #165
• Topic/upstream session lifecycle 20260412 by @everett4320 in #167
• Revert "Topic/upstream session lifecycle 20260412" by @bbsngg in #169
• feat: add /btw side-question support for Gemini and Codex providers by @davidliuk in #171
• Add reader modes and chat-area file overlay by @bbsngg in #173
• Fix ResearchLab Run Next Task prompt injection by @bbsngg in #178
• feat: multi-session tab management with LRU cache and navigation source tracking by @liuyixin-louis in #174
• feat(chat): add copy, resend, and edit actions by @r1ng13 in #161
• fix: address multi-tab PR review follow-ups by @liuyixin-louis in #181
• feat(md): Think/Research mode auto-save, hyperlinks & skill integration by @lichao-sun in #166
• Feat/compute warning skill by @zli12321 in #160
• feat: add GPT-5.5 and Opus 4.7 chat models by @bbsngg in #198
• Feat/news feed GitHub hf wechat by @zli12321 in #199
• Fix Codex CLI version drift in embedded shell by @bbsngg in #200
• Fix inno figure provider key handling by @bbsngg in #197
• Refresh model lists: add Claude Fable 5 and Gemini 3.5 Flash, update provider defaults by @bbsngg in #205
• chore: update agent runtime versions by @bbsngg in #206
• fix typos in skills/distributed-training/deepspeed/SKILL.md by @kevgeoleo in #208
• Support non-ASCII project paths (Claude session listing + Codex header crash) by @bbsngg in #207
• feat: support GPT-5.6 reasoning efforts by @bbsngg in #209
• fix: quote argument-hint YAML values so Copilot CLI ≥1.0.65 loads all skills (closes #210) by @thejesh23 in #211
• fix(perf): stop re-parsing the whole Codex session history on every refresh by @davidliuk in #212
• docs: add EMNLP 2026 paper info by @bbsngg in #217
• feat(skills): add writing-style skill for prose rule application by @MarkZhang777 in #196
• feat(models): discover the harness's own model list instead of hardcoding it by @davidliuk in #213
• Feat/opus5 support by @MarkZhang777 in #216
• fix: drop stray node_modules symlink and quote autoresearch frontmatter by @bbsngg in #218
• perf(client): defer workspace and heavy UI bundles by @davidliuk in #220
• fix(models): per-provider discovery state, Fable 5.1 in the picker, Codex effort from the CLI by @bbsngg in #219
• feat(models): discover Gemini models from the API; keep the Gemini CLI usable on API keys by @bbsngg in #221
• fix(codex): update bundled CLI and SDK to 0.154.0 by @MarkZhang777 in #223
• fix(auth): drop the static JWT secret fallback by @bbsngg in #226
• fix(auth): keep the sign-in option reachable on first run by @bbsngg in #227
• fix(codex): discover models from the bundled CLI the chat actually runs on by @bbsngg in #224
• fix: complete loading review follow-ups and isolate lazy retries by @davidliuk in #222
• feat: macOS Electron native titlebar, dock icon, and drag regions by @davidliuk in #190
• fix(electron): keep collapsed header clear of macOS window buttons by @bbsngg in #231
• docs: add Parallel Search MCP setup for Claude sessions by @georgeatparallel in #233
• feat(local): support vLLM and SGLang OpenAI-compatible servers by @davidliuk in #228
• Add deed.page MCP guide and protect MCP credential logs by @f-o-x11 in #234
• feat(aris): sync self-evolution Workflow M into Auto Research Hub by @davidliuk in #229
• Release 1.2.0: redesigned research workbench by @bbsngg in #239
• fix: desktop reopen crash, native ABI check, skill access, session switching, Codex CLI policy by @bbsngg in #238
• fix(ci): build the Windows desktop installer on windows-2022 by @bbsngg in #240

New Contributors

• @lichao-sun made their first contribution in #166
• @kevgeoleo made their first contribution in #208
• @thejesh23 made their first contribution in #211
• @georgeatparallel made their first contribution in #233
• @f-o-x11 made their first contribution in #234

Full Changelog: v1.1.4...v1.2.0