docs(release): add final v0.8.29 late fixes

CHANGELOG.md

@@ -9,7 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 A maintenance release anchored by a regression fix for the
 "scroll demon" (#1085 class, re-introduced by v0.8.27's flicker
-patch) and a wrong-project session-restore bug (#1395). Plus 16
+patch) and a wrong-project session-restore bug (#1395). Plus 20
 community PRs covering MCP transport, prompt steering, auto-routing
 language coverage, web-search SERP filtering, and broad test
 coverage additions.
@@ -72,6 +72,24 @@ coverage additions.
   previously the project context's inferred `lang` could override
   the latest user message, leading to English thinking for a
   Chinese turn.
+- **Deferred tools hydrate their schema before first execution**
+  (#1419, PR #1429 from **@SamhandsomeLee**). When the model asks
+  for a deferred tool such as `edit_file` before seeing its schema,
+  the engine now loads the tool, returns a non-executed hydration
+  result with the expected fields, and requires a retry instead of
+  executing guessed argument names. Common `edit_file` aliases such
+  as `old_string -> search` and `new_string -> replace` are called
+  out in the retry hint.
+- **DeepSeek public aliases replay thinking-mode tool turns**
+  (PR #1428 from **@Beltran12138**). `deepseek-chat` and
+  `deepseek-reasoner` now classify as V4 reasoning models for
+  `reasoning_content` replay, preventing second-turn HTTP 400s
+  after tool calls when users keep the onboarding default model
+  alias.
+- **Skill completions no longer flood the top-level slash menu**
+  (#1437, PR #1442 from **@reidliu41**). Installed skills now
+  complete under `/skill <name>` while the root `/` menu stays
+  focused on built-in commands.
 
 ### Added
 
@@ -131,6 +149,10 @@ coverage additions.
   (`header_renders_version_chip_when_width_allows` and
   `narrow_header_drops_version_chip_before_dropping_mode`)
   pinning the version chip's cascade priority.
+- **Workspace/session test isolation** tightened (PR #1431 from
+  **@reidliu41**). Git-root detection ignores invalid parent `.git`
+  markers, env-mutating tests share the crate-wide test lock, and
+  the streamable HTTP MCP mock server stays alive for the full test.
 
 ## [0.8.28] - 2026-05-10
 

README.md

@@ -228,7 +228,7 @@ deepseek --provider ollama --model deepseek-coder:1.3b
 ## What's New In v0.8.29
 
 A maintenance release anchored by a v0.8.27 / v0.8.28 regression fix
-plus 16 community PRs. [Full changelog](CHANGELOG.md).
+plus 20 community PRs. [Full changelog](CHANGELOG.md).
 
 - **Scroll demon, gone for good** (#1085 regression). Parallel sub-
   agents running `exec_shell` would scroll the alt-screen out from
@@ -284,6 +284,18 @@ plus 16 community PRs. [Full changelog](CHANGELOG.md).
   reasoning-effort picker correctly route Chinese / Japanese
   technical queries instead of falling through to the generic
   baseline.
+- **Deferred tools hydrate schemas before first execution** (#1419,
+  PR #1429 from **@SamhandsomeLee**) — `edit_file` and other
+  deferred tools now load, show their expected fields, and ask the
+  model to retry instead of executing guessed argument names.
+- **DeepSeek aliases replay thinking-mode tool turns** (PR #1428
+  from **@Beltran12138**) — `deepseek-chat` and
+  `deepseek-reasoner` now get the same `reasoning_content` replay
+  treatment as explicit V4 model IDs, avoiding second-turn 400s
+  after tool calls.
+- **Skill completions stay under `/skill`** (#1437, PR #1442 from
+  **@reidliu41**) — large local skill collections no longer crowd
+  the root slash-command menu.
 - **Sync-to-CNB workflow hardened** — explicit `permissions:
   contents: read`, narrowed trigger to `main` + `v*` tags (no longer
   mirrors feature branches), `actions/checkout` bumped v3 → v4.
@@ -293,7 +305,8 @@ plus 16 community PRs. [Full changelog](CHANGELOG.md).
   from **@linzhiqin2003**).
 
 Thanks to **@linzhiqin2003** (10 landings this cycle),
-**@reidliu41** (3 landings), **@Wenjunyun123**, **@hlx98007**,
+**@reidliu41** (5 landings), **@SamhandsomeLee**,
+**@Beltran12138**, **@Wenjunyun123**, **@hlx98007**,
 **@Liu-Vince**, and **@shenxiaodaosanhua** for the bug report.
 
 ---

README.zh-CN.md

@@ -196,7 +196,7 @@ deepseek --provider ollama --model deepseek-coder:1.3b
 
 维护版本，核心是修复 v0.8.27 / v0.8.28 引入的"滚动幽灵"回归
 （#1085 类问题）和 Ctrl+R 会话恢复跨项目泄漏的问题（#1395），
-外加 16 个社区 PR。[完整更新日志](CHANGELOG.md)。
+外加 20 个社区 PR。[完整更新日志](CHANGELOG.md)。
 
 - **"滚动幽灵"彻底修复**（#1085 回归）。并行子代理运行
   `exec_shell` 时，alt-screen 会被滚动出 ratatui 差分渲染器的
@@ -244,6 +244,15 @@ deepseek --provider ollama --model deepseek-coder:1.3b
 - **Auto 路由识别 CJK 调试 / 搜索关键词**（PR #1401、#1402，
   来自 **@linzhiqin2003**）— `--model auto` 和推理强度选择器
   现在能正确路由中文 / 日文技术查询，此前会回退到通用基准。
+- **Deferred tools 首次执行前会先加载 schema**（#1419，PR #1429，
+  来自 **@SamhandsomeLee**）— `edit_file` 等延迟加载工具现在会先
+  展示期望字段并要求模型重试，而不是执行模型猜测出来的参数名。
+- **DeepSeek 公开别名会正确回放 thinking-mode 工具轮次**（PR #1428，
+  来自 **@Beltran12138**）— `deepseek-chat` 和
+  `deepseek-reasoner` 现在与显式 V4 模型 ID 一样触发
+  `reasoning_content` replay，避免工具调用后的第二轮 400。
+- **技能补全收敛到 `/skill` 下**（#1437，PR #1442，来自
+  **@reidliu41**）— 本地技能很多时不会再挤满根级 `/` 命令菜单。
 - **`sync-cnb.yml` 工作流加固** — 显式 `permissions: contents:
   read`、`actions/checkout` v3 → v4、触发器收紧到 `main` +
   `v*` 标签（不再镜像 feature 分支）。
@@ -251,9 +260,10 @@ deepseek --provider ollama --model deepseek-coder:1.3b
   `parse_pages_arg`、Web 搜索优先级、`sanitize_stream_chunk`
   控制字节过滤（PR #1403–#1406，来自 **@linzhiqin2003**）。
 
-感谢本周期落地 10 个 PR 的 **@linzhiqin2003**、落地 3 个 PR 的
-**@reidliu41**，以及 **@Wenjunyun123**、**@hlx98007**、
-**@Liu-Vince**，和报告 #1395 的 **@shenxiaodaosanhua**。
+感谢本周期落地 10 个 PR 的 **@linzhiqin2003**、落地 5 个 PR 的
+**@reidliu41**，以及 **@SamhandsomeLee**、**@Beltran12138**、
+**@Wenjunyun123**、**@hlx98007**、**@Liu-Vince**，和报告 #1395 的
+**@shenxiaodaosanhua**。
 
 ---