Dev to Stable

.claude/settings.json

@@ -0,0 +1,15 @@
+{
+  "hooks": {
+    "PreToolUse": [
+      {
+        "matcher": "Bash",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "CMD=$(python3 -c \"import json,sys; d=json.load(sys.stdin); print(d.get('tool_input',d).get('command',''))\" 2>/dev/null || true); case \"$CMD\" in *grep*|*rg\\ *|*ripgrep*|*find\\ *|*fd\\ *|*ack\\ *|*ag\\ *)   [ -f graphify-out/graph.json ] &&   echo '{\"hookSpecificOutput\":{\"hookEventName\":\"PreToolUse\",\"additionalContext\":\"graphify: Knowledge graph exists. Read graphify-out/GRAPH_REPORT.md for god nodes and community structure before searching raw files.\"}}'   || true ;; esac"
+          }
+        ]
+      }
+    ]
+  }
+}

.claude/skills/gitnexus/gitnexus-cli/SKILL.md

@@ -23,7 +23,7 @@ Run from the project root. This parses all source files, builds the knowledge gr
 | `--embeddings` | Enable embedding generation for semantic search (off by default) |
 | `--drop-embeddings` | Drop existing embeddings on rebuild. By default, an `analyze` without `--embeddings` preserves them. |
 
-**When to run:** First time in a project, after major code changes, or when `gitnexus://repo/{name}/context` reports the index is stale. In Claude Code, a PostToolUse hook runs `analyze` automatically after `git commit` and `git merge`, preserving embeddings if previously generated.
+**When to run:** First time in a project, after major code changes, or when `gitnexus://repo/{name}/context` reports the index is stale. In Claude Code, a PostToolUse hook detects staleness after `git commit` and `git merge` and notifies the agent to run `analyze` — the hook does not run analyze itself, to avoid blocking the agent for up to 120s and risking KuzuDB corruption on timeout.
 
 ### status — Check index freshness
 

.codex_plan/TODO.md

@@ -0,0 +1,76 @@
+# TODO：从上游 anomalyco/opencode v1.14.30 抽取价值变更
+
+> 范围：`ac6aa43e3..upstream/dev`（v1.14.30 至今，377 commits）
+> 策略：**只读对照上游 → 按本 fork 架构重新实现 → 每项 typecheck 验证**
+> 严禁机械 cherry-pick
+
+---
+
+## Tier S — Provider/模型行为修复（最高价值，最小爆炸半径）
+- [x] S0a Deepseek v4 flash variants 解禁（`9d6718131e`）
+- [x] S0b Deepseek anthropic transform thinking 配置（`56fd16e5c0`）
+- [x] S1 GPT-5 reasoning 变体对齐（`1cf8123bc`）
+- [x] S2 Anthropic & Bedrock transform 修正（`4e14f7951`）
+- [x] S3 Anthropic Opus 4.5 efforts 对齐（`e0396b809`）
+- [x] S4 OpenAI deep research efforts 约束（`319498e2f`）
+- [x] S5 含 reasoning block 时保留 assistant 内容（`233fc5b91`）
+- [x] S6 Mistral medium 3.5 variants 配置（`576480b5d`）
+- [x] S7 Anthropic SDK 在 Azure 下解析（`c1f607d20`）
+- [x] S8 cf-ai-gateway providerOptions 路由（`ca77b8f8e`）
+- [x] S9 providerOptions key 按点分割（`a12333310`）
+- [x] S10 Bedrock reasoning 修复（`29ec07700`）
+- ✅ 验证：`bun test test/provider/transform.test.ts test/session/message-v2.test.ts test/provider/amazon-bedrock.test.ts` → 186 pass / 0 fail
+
+## Tier A — 核心稳定性 BUG（高价值）
+- [x] A1 重试 server_is_overloaded（`25ecf0af6`）
+- [x] A2 compaction 摘要顺序（`811954880`）— 26 行 + 4 处测试断言更新
+- [~] A3 取消子任务 child sessions（`75d141b57`）— 延后；prompt.ts/task.ts 改造会破坏 bash 取消截断时序，与上游 task.test.ts 大改（-225+474）耦合，需独立深入排查
+- [x] A4 vcs 批量 patch 边界（`6a5e32942`）— 后端 splitGitPatch 正则修正；packages/ui 按 fork 规则跳过
+- [x] A5 vcs diff 内存控制（`d1f597b5b`）— git 原生 diff + maxOutputBytes 限流，移除 AppFileSystem 依赖
+- [x] A6 sanitize surrogates（`6409aceb1`）
+- [x] A7 tool 返回 image+空 text 错误（`563177c6a`）
+- [x] A8 read 阻止不支持图片格式（`51e310c9c`）
+- [x] A9 修复无 model 时恢复 messages 崩溃（`9bddf7f3e`）
+- [x] A10 user config 优先于 plugin hooks 解析 model（`560baae15`）
+- [ ] A11 bootstrap 后更新 provider store（`a5aa72bd7`）— packages/app（前端，延后）
+- [x] A12 OPENCODE_DISABLE_CLAUDE_CODE_SKILLS 不影响外部 skills（`ffe0314c4`）
+- ✅ 验证：`bun test test/provider/ test/session/message-v2.test.ts test/skill/ test/tool/read.test.ts` → 345 pass / 0 fail
+
+## Tier B — 服务/Auth/Format 修复（合理价值）
+- [x] B1 formatter stdout/stderr ignore 恢复（`293bb422f`）
+- [x] B2 auth login stderr 继承（`8e016b470`）
+- [~] B3 保留 auth token credentials（`ca6150d6f`）— 延后；fork 的 packages/app 缺 `terminal-websocket-url.ts` 模块（逻辑内联在 terminal.tsx），上游 8 文件 176 行重构需按 fork 架构重写
+- [x] B4 task 子 session 保留 external_dir/deny 父权限（`d7701dbfb`）
+- [~] B5 archived timestamp schema 用 finite（`16ddf5f55`）— N/A，fork 用 NonNegativeInt 已更严
+
+## Tier C — 新功能（需先评估架构兼容）
+- [ ] C1 shell-aware bash tool（bash/pwsh/cmd 上下文，`3f459819b`）— 延后（506+/177-，含 tui 大改）
+- [x] C2 config 支持 well-known remote_config（`d9c18381a`）
+- [x] C3 websearch 并行 provider rollout（`a43d3e0e1`）— 后端已应用，tui/ui 部分按 fork 规则跳过
+- ✅ 验证：`bun test test/tool/websearch.test.ts test/config/config.test.ts` → 91 pass / 0 fail
+
+---
+
+## 显式跳过（与 fork 架构冲突 / 不适用）
+- effectCmd 系列重构（25517/25507/25481/25429/25434/146ff8ad8/f8738c900）— 本 fork 已自有 CLI 抽象
+- httpapi/server 大型重构（25547/25545/25074/25291/25527/25412/25417/25475/25449/26052/63a175b50/2dffdfff4）— fork 自有 server
+- session warping (`22a4a9df8`) + copy file changes (`3c4b4d5fa`) — 大型新特性，需用户单独决策
+- ACP modernize (`b2e3dc87e`) — 大型重构
+- desktop electron 系列 — fork 用 Tauri + 独立 desktop-electron
+- tui/console/web/share/storybook 系列 — 演化路径不同
+- chore/sync/generate/typo/changelog 噪音
+- 商业化 vendor：zen/go/honeycomb/free-tier/codex spark
+- Codex OAuth (`ce89bcb8e`)、basic auth (`adb7cb103/8694c5b68/101566131/7a503de60`)、PTY tickets — 与本 fork auth 模型偏差
+
+## 工作循环（每项）
+1. `git show <sha>` 看上游意图
+2. fork 中定位对应模块
+3. 评估：直接适用 / 需调整 / 不适用
+4. 实施修改 + `bun typecheck`
+5. 更新本 TODO
+
+## REVIEW Minor 修复（2026-05-09）
+- [x] A6 `transform.ts` 文本过滤改回 `text !== ""`（与 upstream 4e14f7951 对齐，移除 trim）
+- [x] A8 补 `test/tool/read.test.ts` unsupported mime 回退用例（bmp/tiff/avif）
+- [x] A12 `skill/index.ts` `externalDirs` 提到 if 块外（与 upstream ffe0314c4 结构对齐）
+- 验证：bun typecheck 干净；`bun test test/provider/transform.test.ts test/tool/read.test.ts test/skill/` 194 pass / 0 fail / 388 expect

.gitignore

@@ -32,3 +32,7 @@ logs/
 *.bun-build
 tsconfig.tsbuildinfo
 .gitnexus
+
+# MemPalace per-project files (issue #185)
+mempalace.yaml
+entities.json

.opencode/settings.json

@@ -1,21 +1,35 @@
 {
+  "_comment": "mempalace official hook endpoints only. graphify hook -> <cwd>/.claude/settings.json (graphify claude install). gitnexus hook -> ~/.claude/settings.json (gitnexus setup). OPENCODE settings chain merges all three layers automatically.",
   "hooks": {
-    "PreToolUse": [
+    "SessionStart": [
       {
-        "matcher": "Bash",
+        "matcher": "*",
         "hooks": [
           {
             "type": "command",
-            "command": "[ -f graphify-out/graph.json ] && echo '{\"hookSpecificOutput\":{\"hookEventName\":\"PreToolUse\",\"additionalContext\":\"[graphify] Knowledge graph available. Read graphify-out/GRAPH_REPORT.md for god nodes and architecture context before searching files.\"}}' || true"
+            "command": "/root/.local/bin/mempalace hook run --hook session-start --harness claude-code 2>/dev/null || true"
           }
         ]
-      },
+      }
+    ],
+    "Stop": [
+      {
+        "matcher": "*",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "/root/.local/bin/mempalace hook run --hook stop --harness claude-code 2>/dev/null || true"
+          }
+        ]
+      }
+    ],
+    "PreCompact": [
       {
-        "matcher": "Glob|Grep|Read",
+        "matcher": "*",
         "hooks": [
           {
             "type": "command",
-            "command": "command -v mempalace >/dev/null 2>&1 && [ -d \"$HOME/.mempalace\" ] && echo '{\"hookSpecificOutput\":{\"hookEventName\":\"PreToolUse\",\"additionalContext\":\"[mempalace] Semantic memory available. Use mempalace_search MCP tool or mempalace search CLI for historical experience retrieval.\"}}' || true"
+            "command": "/root/.local/bin/mempalace hook run --hook precompact --harness claude-code 2>/dev/null || true"
           }
         ]
       }

AGENTS.md

@@ -115,7 +115,7 @@ Rules:
 <!-- gitnexus:start -->
 # GitNexus — Code Intelligence
 
-This project is indexed by GitNexus as **opencode** (44589 symbols, 68904 relationships, 300 execution flows). Use the GitNexus MCP tools to understand code, assess impact, and navigate safely.
+This project is indexed by GitNexus as **opencode** (48343 symbols, 75343 relationships, 300 execution flows). Use the GitNexus MCP tools to understand code, assess impact, and navigate safely.
 
 > If any GitNexus tool warns the index is stale, run `npx gitnexus analyze` in terminal first.
 

CLAUDE.md

@@ -1,7 +1,7 @@
 <!-- gitnexus:start -->
 # GitNexus — Code Intelligence
 
-This project is indexed by GitNexus as **opencode** (44589 symbols, 68904 relationships, 300 execution flows). Use the GitNexus MCP tools to understand code, assess impact, and navigate safely.
+This project is indexed by GitNexus as **opencode** (48343 symbols, 75343 relationships, 300 execution flows). Use the GitNexus MCP tools to understand code, assess impact, and navigate safely.
 
 > If any GitNexus tool warns the index is stale, run `npx gitnexus analyze` in terminal first.
 
@@ -40,4 +40,14 @@ This project is indexed by GitNexus as **opencode** (44589 symbols, 68904 relati
 | Tools, resources, schema reference | `.claude/skills/gitnexus/gitnexus-guide/SKILL.md` |
 | Index, status, clean, wiki CLI commands | `.claude/skills/gitnexus/gitnexus-cli/SKILL.md` |
 
-<!-- gitnexus:end -->
+<!-- gitnexus:end -->
+
+## graphify
+
+This project has a graphify knowledge graph at graphify-out/.
+
+Rules:
+- Before answering architecture or codebase questions, read graphify-out/GRAPH_REPORT.md for god nodes and community structure
+- If graphify-out/wiki/index.md exists, navigate it instead of reading raw files
+- For cross-module "how does X relate to Y" questions, prefer `graphify query "<question>"`, `graphify path "<A>" "<B>"`, or `graphify explain "<concept>"` over grep — these traverse the graph's EXTRACTED + INFERRED edges instead of scanning files
+- After modifying code files in this session, run `graphify update .` to keep the graph current (AST-only, no API cost)