• docs: acknowledge linux do community (bfee651)
• docs: highlight kimi open source partner (1277760)

Full Changelog: v1.4.1...v1.5.0

Staging release candidate for InkOS v1.5.0.

Status:

• CI test matrix passed on Node 20 / 22 / 24.
• Smoke test passed: CLI help/version, core import, tarball workspace dependency check.
• npm canary publish passed and the canary install was verified by CI.

Use this tag for source-level and canary-package verification before the formal v1.5.0 release. Formal npm/latest publishing will run from the final v1.5.0 tag.

• chore(release): bump version to 1.4.1 (429b5b6)
• feat(core): make long-form review retries configurable (c997dc5)
• docs: refresh zh readme for InkOS Short (811f280)
• fix(core): keep MiniMax fix scoped to default provider (7300a41)
• docs: align readmes with v1.4 writing flow (1209405)
• fix(core): route MiniMax through OpenAI-compatible endpoint (8af863e)
• perf(core): lighten long-form write cycle (92845cc)
• fix(core): normalize short artifact paths (3e05651)

Full Changelog: v1.4.0...v1.4.1

• docs(skill): add kkaiapi localized links (126ff39)
• docs: refine readme release highlights (947bd6f)
• docs: mention aggregator access in 1.4 intro (a91d624)
• docs: fold aggregator note into short intro (7f7548d)
• docs: move aggregator note to readme intro (a6f2edd)
• docs: keep aggregator provider links neutral (2811c39)
• docs(studio): clarify aggregator provider positioning (3cd2f42)
• docs: add aggregator provider links (a153ead)
• chore(release): prepare 1.4.0 (fa6ef92)
• test(cli): relax tui dashboard render timeout (2a10ade)
• fix(core): avoid length normalizer output truncation (f339e83)
• feat(studio): handle direct content edits before agent (1f56c16)
• docs(agent): support chat cover prompt revisions (11237aa)
• docs(skill): document standalone cover tool (0ea2d10)
• fix(studio): preserve generated artifact details (a21ceca)
• fix(studio): persist project chat sessions (046dc60)
• refactor(cli): move short command to public fiction module (80555be)
• refactor(short): use public short fiction naming (1c48fbe)
• docs(readme): add short fiction and cover workflow (231294c)
• fix(studio): show generated cover previews inline (c59e78d)
• feat(studio): add standalone cover generation tool (b2185eb)
• Revert "fix(studio): restore orphan chat sessions on reload" (1f0ba33)
• fix(studio): restore orphan chat sessions on reload (49feeff)
• feat(studio): render generated short covers (6331be6)
• fix(short): use images endpoint for cover providers (4d0fe67)
• fix(agent): keep cover failures factual (4ab2c4d)
• fix(short): remove gpt-5.5 from cover config (981d78f)
• feat(short): add cover generation config (7ef5a65)
• fix(studio): refresh llm mirror after service switch (9ca5898)
• fix(short): recover chapters from repeated title tags (c2e8c95)
• fix(core): repair runtime hooks and cap legacy context (86956f8)
• fix(core): guard length normalizer truncation (0f1dee5)
• refactor(short): centralize fiction prompts (a081ff6)
• feat(studio): let chat run short fiction pipeline (9a7ee95)
• chore(short): use neutral public workflow wording (6288215)
• fix(short): avoid exposing packaging constraints in prompts (06f7250)
• fix(short): require API key for cover generation (61259bb)
• feat(short): generate cover image in public run (3ac686c)
• feat(short): add benchmark-free public run chain (d3b1562)
• fix(studio): simplify aggregator service list (fdb86a0)

Full Changelog: v1.3.12...v1.4.0

• chore(release): prepare v1.3.12 (650ad4b)
• feat(studio): add service quick links (be24fdd)
• docs: refresh Chinese community QR (54f752e)
• docs: restore kkaiapi release links (12a20de)
• docs: remove kkaiapi release wording (3941664)
• docs: link kkaiapi in readmes (5925b81)

Full Changelog: v1.3.11...v1.3.12

• chore(release): prepare v1.3.11 (bf56ae4)
• feat(llm): add kkaiapi and harden service setup (684367f)
• fix(core): honor requested story line proportions (603dc1b)
• fix(core): sanitize generated hook ids (dfa88aa)
• feat(studio): route book creation through chat (25f9e06)
• Update README.md (7f6e80a)
• Add files via upload (36e68c1)
• fix(core): lazy-load notification transports (b08d59b)
• fix(studio): prefer live Volcengine probe models (3d4a4e1)
• fix(studio): surface create config errors (d597ce4)
• fix(studio): localize runtime error copy (f0ade5f)
• fix(studio): simplify book creation form (d1566db)

Full Changelog: v1.3.10...v1.3.11

• chore(release): prepare v1.3.10 (601ce06)
• fix(create): normalize book platform aliases (690a62f)
• fix(agent): normalize sub-agent platform args (5072000)

Full Changelog: v1.3.9...v1.3.10

• chore(release): prepare v1.3.9 (446e063)
• fix(studio): restore book settings route (7b2dc11)
• fix(studio): keep book creation on an orphan session (686dbf1)

Full Changelog: v1.3.8...v1.3.9

• test(llm): cover empty-key local model probing (b2405e3)
• chore(release): prepare v1.3.8 (d18c484)
• fix(llm): support local model creation flows (deceab8)
• fix(studio): route book creation to dedicated page (c0788d0)
• feat(llm): add Kimi Code provider support (da0c24f)

Full Changelog: v1.3.7...v1.3.8

Release Focus

长篇写作质量收紧:把近期验证过的网文写法规则落到 Writer、Planner、Architect 与后置校验中,重点改善开篇抓人、章节密度、伏笔兑现、段落节奏和架构稿完整性。

Improvements

• Writer prompt 新增看点密度、移动端段落、开篇第一屏、章节断章和人物行动动机等写作约束,减少空转铺垫和报告式正文。
• Planner / Architect 更明确地承接黄金开篇、章节目标、hook 账和段落式 foundation 输出要求。
• Hook ledger 要求 advance / resolve 项在正文里有可定位的动作、物件、对话或事件兑现,减少"账本里有、正文里没有"的断层。
• 段落密度规则收紧:密度来自语义和场景推进,不是把正文切成电报体;连续短段会被后置规则识别。

Bug Fixes

• 修复 Architect 在扩展输出时可能漏掉 5 个 foundation SECTION 块的问题。
• 修复 hook ledger payoff 检查过于宽松,导致侧面暗示也可能被误判为兑现的问题。
• 修复写作 prompt 对段落尺寸描述不够明确,模型容易在 1-3 点密度规则下过度碎段的问题。

Verification

• GitHub Actions Release workflow passed on Node 20 and Node 22.
• Smoke test passed: CLI help/version, core import, tarball workspace dependency check.
• Canary publish and installation verification passed.
• Latest publish and installation verification passed.