[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EHTTCK14G47KAJA31M2QPZ":3,"skill-same:01M7EHTTCK14G47KAJA31M2QPZ":44},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":23,"commitSha":24,"skillDirectory":25,"forksCount":26,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M7EHTTCK14G47KAJA31M2QPZ","code-nav",null,"使用仓库已有的 Pyright language server 做 Python 跨文件语义导航。用于改函数前确认调用者\u002F被调用者、收敛 grep 结果到真实语义引用、检查符号爆炸半径，或在陌生模块里追踪引用链。必须通过 readiness gate 才接受语义结果；把 LSP semantic scope 与 repo-visible lexical candidates 分开报告；遇到符号歧义、协议错误、server 提前退出或未就绪时 fail-closed。常用入口：scripts\u002Fcode_nav.py doctor \u002F refs \u002F callers \u002F callees。","deyi2026\u002Fllm-first-loop","deyi2026","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F267286635?v=4","01M7EHT3KGDPR2R5HCCRQA3HYM","01M7EHT3M140G7BHY5CBAJVRH4",1,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Fskills\u002Fcode-nav\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: code-nav\ndescription: 使用仓库已有的 Pyright language server 做 Python 跨文件语义导航。用于改函数前确认调用者\u002F被调用者、收敛 grep 结果到真实语义引用、检查符号爆炸半径，或在陌生模块里追踪引用链。必须通过 readiness gate 才接受语义结果；把 LSP semantic scope 与 repo-visible lexical candidates 分开报告；遇到符号歧义、协议错误、server 提前退出或未就绪时 fail-closed。常用入口：scripts\u002Fcode_nav.py doctor \u002F refs \u002F callers \u002F callees。\n---\n\n# code-nav — 按需 Pyright 语义导航\n\n## 核心原则\n\n1. **先证明 ready，再信结果。** `initialize -> initialized -> didOpen -> publishDiagnostics(target)` 是当前 Pyright 1.1.411 已验证的 readiness contract。未收到目标文件 diagnostics 时命令必须失败并返回非零；不要采用部分结果。\n2. **区分 semantic 与 lexical。** LSP 结果只声明在当前 Pyright program\u002Fconfig scope 内成立。`refs` 默认另跑 repo-visible lexical cross-check；这些文本命中只能叫 lexical candidates，不能直接叫“Pyright 漏掉的 references”。\n3. **调用点和定义点分开。** `callers`\u002F`callees` 输出 `site_path\u002Fsite_lines` 与 definition\u002Ftarget path；必须保留全部 `fromRanges`，不要只看第一个调用点。\n4. **歧义就失败。** 同文件多个同名 definition、或 `prepareCallHierarchy` 返回多个 item 时，不允许自动选第一个。\n5. **路径以 repo root 为准。** 从仓库根或子目录执行应得到同一 semantic 结果；目标文件必须位于 repo 内且为 Python 文件。\n\n## 运行依赖\n\n`code_nav.py` 的 Python 实现不新增第三方包，但运行依赖：\n\n- `git` CLI；\n- 已安装的 `pyright-langserver`。\n\nmanaged worktree 通常没有自己的 `.venv`。脚本会按以下顺序找 language server：\n\n1. 当前 code root 的 `.venv\u002Fbin\u002Fpyright-langserver`；\n2. Git common root 的共享 `.venv\u002Fbin\u002Fpyright-langserver`；\n3. `PATH`。\n\n不要在 Skill 里假设 `$PWD\u002F.venv` 一定存在。\n\n## 用法\n\n```bash\n# 0) 自检：server \u002F version \u002F config \u002F capability \u002F readiness\npython3 scripts\u002Fcode_nav.py doctor\n\n# 1) 谁调用它\npython3 scripts\u002Fcode_nav.py callers src\u002Fllm_loop\u002Fcore\u002Frun_context.py workspace_base\n\n# 2) 它调用谁\npython3 scripts\u002Fcode_nav.py callees src\u002Fllm_loop\u002Ffactory.py build_engine\n\n# 3) 语义引用 + repo-visible lexical complement\npython3 scripts\u002Fcode_nav.py refs src\u002Fllm_loop\u002Fcore\u002Fmessage.py ToolResult --limit 20\n\n# 4) 机器可读\npython3 scripts\u002Fcode_nav.py refs src\u002Fllm_loop\u002Fcore\u002Fmessage.py ToolResult --json\n```\n\n退出码：\n\n- `0`：readiness\u002Fcapability\u002Fambiguity gate 后成功，且有 semantic result；\n- `1`：参数、环境、协议、readiness、ambiguity、repo lock 或默认 cross-check 失败；\n- `2`：readiness 已确认且查询正常完成，但当前 semantic scope 内结果为 0。\n\n`rc=2` **不能**解释成 repo-wide absence。\n\n## 机器输出怎么读\n\n成功 JSON 使用 `schema=lfl.code_nav.v1`，重点看：\n\n- `status`: `ok` \u002F `no_results` \u002F `not_ready` 等；\n- `trust.analysis_ready`；\n- `trust.readiness_signal`；\n- `trust.semantic_scope`；\n- `trust.repo_lexical_cross_check`（`refs`）；\n- `cross_check.universe=git-visible-python`；\n- `cross_check.occurrences \u002F matching_lines \u002F matching_files`；\n- `inside_configured_include` \u002F `outside_configured_include`。\n\nlexical universe 来自：\n\n```bash\ngit ls-files -co --exclude-standard -- '*.py'\n```\n\n即 tracked Python + non-ignored untracked Python。普通 ignored 临时\u002F生成文件不进入统计。\n\n## 2026-09-28 hardened 基线\n\n环境：production-based hardening worktree，base gen159 `889634f21...`，Pyright `1.1.411`。\n\n| 操作 | 当前观测 |\n|---|---|\n| `doctor` | `status=ok`、ready=true、`langserver_source=git_common_root_venv` |\n| `refs ToolResult` | **729 semantic locations \u002F 82 files** |\n| root vs `scripts\u002F` 子目录执行 | semantic locations 完全一致 |\n| lexical complement | **153 occurrences \u002F 153 lines \u002F 47 files** |\n| configured include=`src` 内 lexical-only | 3 occurrences \u002F 3 files |\n| configured include 外 lexical-only | **150 occurrences \u002F 44 files** |\n| `callers workspace_base` | 21 callers |\n| `callees build_engine` | 361 outgoing entries；多 `fromRanges` 完整保留 |\n| `--ready-timeout 0` | rc=1 \u002F `status=not_ready` |\n\n这些数字是**当前代码快照的观测 baseline，不是固定验收门**。代码演进后 exact counts 变化是正常的；必须保留的是 readiness、路径、scope、分类和字段语义。\n\n## 协议硬约束\n\n- `initialize` 成功后必须发送 `initialized`；漏发可表现为静默挂死。\n- server→client 带 `id` 的消息是 request，必须应答。\n- 已支持 request 走 allowlist；未知 request 返回 JSON-RPC `-32601 Method not found`，不要假装 `result:null` 成功。\n- reader framing\u002FJSON 错误或 server 提前退出必须 fail-fast，不等满 query timeout。\n- 正常结束走 `shutdown -> exit`，terminate\u002Fkill 只作兜底。\n- 同一 repo 默认用 advisory lock 限制并发 langserver，避免瞬时 fan-out。\n\n## callers \u002F callees 字段语义\n\n`callers`：\n\n- `site_path\u002Fsite_lines`：caller 内真实调用点；\n- `caller_definition_path\u002Fcaller_definition_line`：caller 定义位置。\n\n`callees`：\n\n- `site_path\u002Fsite_lines`：当前 caller 文件里的调用点；\n- `target_path\u002Ftarget_definition_line`：被调用函数定义位置。\n\n例如真实 hardened canary 已得到：`factory.py:397 -> config.py:731 ensure_dirs()`；这避免了旧原型把 caller 行号错误贴到 callee 文件上的伪定位。\n\n## 已知盲区\n\n1. **动态派发仍解不了**：`getattr`、注册表、字符串 dispatch 等要沿注册结构追踪。\n2. **只覆盖 Python**：TS\u002FJS 尚无对应 LSP wrapper。\n3. **readiness signal 是版本限定 contract**：`publishDiagnostics` 不是 LSP 规范定义的“workspace 全索引完成”通用保证；当前结论基于 Pyright 1.1.411 的 fixture 与真实仓库对照。升级 Pyright 后要重跑真实 fixture\u002Fcanary。\n4. **零常驻有启动成本**：每次调用都会重建 program；不要在高频循环里调用。\n5. **semantic scope 不是 repo 全量真相**：即使 `status=ok`，也只说明当前 Pyright program scope 内的语义结果成功。需要 repo 级排查时同时看 lexical complement。\n\n## 固定核验动作\n\n在高风险修改前：\n\n```bash\npython3 scripts\u002Fcode_nav.py doctor --json\npython3 scripts\u002Fcode_nav.py refs \u003Cfile> \u003Csymbol> --json\npython3 scripts\u002Fcode_nav.py callers \u003Cfile> \u003Csymbol> --json\n```\n\n若 lexical candidates 与 semantic results 不一致，回读对应源文件判断它们是注释\u002F字符串\u002F同名符号，还是 program scope 外的真实使用；不要把文本差额直接升级成语义结论。\n","# code-nav — 按需 Pyright 语义导航\n\n## 核心原则\n\n1. **先证明 ready，再信结果。** `initialize -> initialized -> didOpen -> publishDiagnostics(target)` 是当前 Pyright 1.1.411 已验证的 readiness contract。未收到目标文件 diagnostics 时命令必须失败并返回非零；不要采用部分结果。\n2. **区分 semantic 与 lexical。** LSP 结果只声明在当前 Pyright program\u002Fconfig scope 内成立。`refs` 默认另跑 repo-visible lexical cross-check；这些文本命中只能叫 lexical candidates，不能直接叫“Pyright 漏掉的 references”。\n3. **调用点和定义点分开。** `callers`\u002F`callees` 输出 `site_path\u002Fsite_lines` 与 definition\u002Ftarget path；必须保留全部 `fromRanges`，不要只看第一个调用点。\n4. **歧义就失败。** 同文件多个同名 definition、或 `prepareCallHierarchy` 返回多个 item 时，不允许自动选第一个。\n5. **路径以 repo root 为准。** 从仓库根或子目录执行应得到同一 semantic 结果；目标文件必须位于 repo 内且为 Python 文件。\n\n## 运行依赖\n\n`code_nav.py` 的 Python 实现不新增第三方包，但运行依赖：\n\n- `git` CLI；\n- 已安装的 `pyright-langserver`。\n\nmanaged worktree 通常没有自己的 `.venv`。脚本会按以下顺序找 language server：\n\n1. 当前 code root 的 `.venv\u002Fbin\u002Fpyright-langserver`；\n2. Git common root 的共享 `.venv\u002Fbin\u002Fpyright-langserver`；\n3. `PATH`。\n\n不要在 Skill 里假设 `$PWD\u002F.venv` 一定存在。\n\n## 用法\n\n```bash\n# 0) 自检：server \u002F version \u002F config \u002F capability \u002F readiness\npython3 scripts\u002Fcode_nav.py doctor\n\n# 1) 谁调用它\npython3 scripts\u002Fcode_nav.py callers src\u002Fllm_loop\u002Fcore\u002Frun_context.py workspace_base\n\n# 2) 它调用谁\npython3 scripts\u002Fcode_nav.py callees src\u002Fllm_loop\u002Ffactory.py build_engine\n\n# 3) 语义引用 + repo-visible lexical complement\npython3 scripts\u002Fcode_nav.py refs src\u002Fllm_loop\u002Fcore\u002Fmessage.py ToolResult --limit 20\n\n# 4) 机器可读\npython3 scripts\u002Fcode_nav.py refs src\u002Fllm_loop\u002Fcore\u002Fmessage.py ToolResult --json\n```\n\n退出码：\n\n- `0`：readiness\u002Fcapability\u002Fambiguity gate 后成功，且有 semantic result；\n- `1`：参数、环境、协议、readiness、ambiguity、repo lock 或默认 cross-check 失败；\n- `2`：readiness 已确认且查询正常完成，但当前 semantic scope 内结果为 0。\n\n`rc=2` **不能**解释成 repo-wide absence。\n\n## 机器输出怎么读\n\n成功 JSON 使用 `schema=lfl.code_nav.v1`，重点看：\n\n- `status`: `ok` \u002F `no_results` \u002F `not_ready` 等；\n- `trust.analysis_ready`；\n- `trust.readiness_signal`；\n- `trust.semantic_scope`；\n- `trust.repo_lexical_cross_check`（`refs`）；\n- `cross_check.universe=git-visible-python`；\n- `cross_check.occurrences \u002F matching_lines \u002F matching_files`；\n- `inside_configured_include` \u002F `outside_configured_include`。\n\nlexical universe 来自：\n\n```bash\ngit ls-files -co --exclude-standard -- '*.py'\n```\n\n即 tracked Python + non-ignored untracked Python。普通 ignored 临时\u002F生成文件不进入统计。\n\n## 2026-09-28 hardened 基线\n\n环境：production-based hardening worktree，base gen159 `889634f21...`，Pyright `1.1.411`。\n\n| 操作 | 当前观测 |\n|---|---|\n| `doctor` | `status=ok`、ready=true、`langserver_source=git_common_root_venv` |\n| `refs ToolResult` | **729 semantic locations \u002F 82 files** |\n| root vs `scripts\u002F` 子目录执行 | semantic locations 完全一致 |\n| lexical complement | **153 occurrences \u002F 153 lines \u002F 47 files** |\n| configured include=`src` 内 lexical-only | 3 occurrences \u002F 3 files |\n| configured include 外 lexical-only | **150 occurrences \u002F 44 files** |\n| `callers workspace_base` | 21 callers |\n| `callees build_engine` | 361 outgoing entries；多 `fromRanges` 完整保留 |\n| `--ready-timeout 0` | rc=1 \u002F `status=not_ready` |\n\n这些数字是**当前代码快照的观测 baseline，不是固定验收门**。代码演进后 exact counts 变化是正常的；必须保留的是 readiness、路径、scope、分类和字段语义。\n\n## 协议硬约束\n\n- `initialize` 成功后必须发送 `initialized`；漏发可表现为静默挂死。\n- server→client 带 `id` 的消息是 request，必须应答。\n- 已支持 request 走 allowlist；未知 request 返回 JSON-RPC `-32601 Method not found`，不要假装 `result:null` 成功。\n- reader framing\u002FJSON 错误或 server 提前退出必须 fail-fast，不等满 query timeout。\n- 正常结束走 `shutdown -> exit`，terminate\u002Fkill 只作兜底。\n- 同一 repo 默认用 advisory lock 限制并发 langserver，避免瞬时 fan-out。\n\n## callers \u002F callees 字段语义\n\n`callers`：\n\n- `site_path\u002Fsite_lines`：caller 内真实调用点；\n- `caller_definition_path\u002Fcaller_definition_line`：caller 定义位置。\n\n`callees`：\n\n- `site_path\u002Fsite_lines`：当前 caller 文件里的调用点；\n- `target_path\u002Ftarget_definition_line`：被调用函数定义位置。\n\n例如真实 hardened canary 已得到：`factory.py:397 -> config.py:731 ensure_dirs()`；这避免了旧原型把 caller 行号错误贴到 callee 文件上的伪定位。\n\n## 已知盲区\n\n1. **动态派发仍解不了**：`getattr`、注册表、字符串 dispatch 等要沿注册结构追踪。\n2. **只覆盖 Python**：TS\u002FJS 尚无对应 LSP wrapper。\n3. **readiness signal 是版本限定 contract**：`publishDiagnostics` 不是 LSP 规范定义的“workspace 全索引完成”通用保证；当前结论基于 Pyright 1.1.411 的 fixture 与真实仓库对照。升级 Pyright 后要重跑真实 fixture\u002Fcanary。\n4. **零常驻有启动成本**：每次调用都会重建 program；不要在高频循环里调用。\n5. **semantic scope 不是 repo 全量真相**：即使 `status=ok`，也只说明当前 Pyright program scope 内的语义结果成功。需要 repo 级排查时同时看 lexical complement。\n\n## 固定核验动作\n\n在高风险修改前：\n\n```bash\npython3 scripts\u002Fcode_nav.py doctor --json\npython3 scripts\u002Fcode_nav.py refs \u003Cfile> \u003Csymbol> --json\npython3 scripts\u002Fcode_nav.py callers \u003Cfile> \u003Csymbol> --json\n```\n\n若 lexical candidates 与 semantic results 不一致，回读对应源文件判断它们是注释\u002F字符串\u002F同名符号，还是 program scope 外的真实使用；不要把文本差额直接升级成语义结论。\n",{"name":5,"description":7},"main","57f220f344f02e6abf6eb55434077d104c209804","skills\u002Fcode-nav",0,"https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop","https:\u002F\u002Fgithub.com\u002Fdeyi2026","2026-11-15 10:23:32",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add deyi2026\u002Fllm-first-loop --skill code-nav",{"tool":36,"command":37},"bunx","bunx skills add deyi2026\u002Fllm-first-loop --skill code-nav",{"tool":39,"command":40},"pnpm","pnpm dlx skills add deyi2026\u002Fllm-first-loop --skill code-nav","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Fskills\u002Fcode-nav\u002FSKILL.md",true,[],[45,52,59,66,73,80],{"skillNo":46,"name":47,"nameZh":6,"description":48,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":49,"categories":50,"keywords":51,"publishedAt":18},"01M7EJ01AM5PZXYNM9GBF6CW12","security-audit","Audit project for security issues.","https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fvendor_gemini_demo\u002Fskills\u002Fsecurity-audit\u002FSKILL.md",[],[],{"skillNo":53,"name":54,"nameZh":6,"description":55,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":56,"categories":57,"keywords":58,"publishedAt":18},"01M7EJ00CD2JXGVB0JETXGEX43","metrics","Collect project metrics.","https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fvendor_codex_demo\u002Fskills\u002Fmetrics\u002FSKILL.md",[],[],{"skillNo":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":18},"01M7EHZZD7DVVTSECGF5HHHQZ5","lint","Lint the repo using project config.","https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fvendor_claude_demo\u002Fskills\u002Flint\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":18},"01M7EHZYGGYKDZDEHVF4YSD22X","report","值班报告生成","https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fstd_plugin_demo\u002Fskills\u002Freport\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":18},"01M7EHZVX0V6ER3X3017K47RK4","deploy","标准部署流程（多行 YAML frontmatter 演示）","https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fstd_plugin_demo\u002Fskills\u002Fdeploy\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M7EHZRENGPKXXFBA3YJ1QF1S","web-self-restart-triage","https:\u002F\u002Fgithub.com\u002Fdeyi2026\u002Fllm-first-loop\u002Fblob\u002Fmain\u002Fskills\u002Fweb-self-restart-triage\u002FSKILL.md",[],[]]