@kairyou/agent-tools 0.10.1 → 0.11.0

package/README.md

@@ -1,6 +1,6 @@
 # Agent Tools
 
-Reusable Agent Skills and installable integrations — statusline, provider usage, and vision — for Codex, Claude Code, and opencode.
+Reusable Agent Skills, plus integrations (statusline, provider usage, vision) that install into Codex, Claude Code, and opencode.
 
 Requires Node.js >= 22.
 
@@ -77,6 +77,7 @@ Runtime capabilities, installed per agent:
 
 ```bash
 npx -y @kairyou/agent-tools@latest <capability> -a <agent...>
+# npx -y github:kairyou/agent-tools <capability> -a <agent...>  # Or the latest code from GitHub (needs Git)
 ```
 
 `--dry-run` previews, `--uninstall` unwires the integration from the agent, and
@@ -218,25 +219,30 @@ To diagnose the provider setup or test recognition quality manually:
 npx -y @kairyou/agent-tools@latest inspect-image <path|url> -q "What are the navbar background color and height?"
 ```
 
-## Run from Git
+## Extras
 
-To run directly from the repository, replace the npm package name with
-`github:kairyou/agent-tools` (Git required):
-
-```bash
-npx -y github:kairyou/agent-tools usage -a codex
-```
-
-Contributors can see the [repository structure](docs/en/repository-structure.md).
+See [extras](docs/en/extras.md) for situational content.
 
 ## FAQ
 
-### Why does global installation fail for PromptScript?
+### Why does installing skills report PromptScript as unsupported?
 
 `PromptScript does not support global skill installation` means that the
 PromptScript agent does not support global installation. It does not affect
 other agents and can be ignored. See [`skills` issue #1352](https://github.com/vercel-labs/skills/issues/1352).
 
+### How do I install skills without GitHub access?
+
+Install from the npm package instead (swap `at-commit` for the skills you want):
+
+```bash
+cd "$(mktemp -d)" && tar -xf "$(npm pack @kairyou/agent-tools --silent)" && npx -y skills@latest add ./package --skill at-commit -g -y
+```
+
+### Want to contribute?
+
+See the [repository structure](docs/en/repository-structure.md).
+
 ## Acknowledgements
 
 - `at-commit` draws on commit-message generation ideas from

package/README.zh-CN.md

@@ -2,7 +2,7 @@
 
 [English](README.md)
 
-可复用的 Agent Skills, 以及面向 Codex, Claude Code 与 opencode 的可安装 integrations: statusline, provider usage 与跨模型识图.
+可复用的 Agent Skills, 以及可装入 Codex, Claude Code 与 opencode 的 integrations: statusline, provider usage 与跨模型识图.
 
 需要 Node.js >= 22.
 
@@ -77,6 +77,7 @@ Runtime capability, 按 agent 选装:
 
 ```bash
 npx -y @kairyou/agent-tools@latest <capability> -a <agent...>
+# npx -y github:kairyou/agent-tools <capability> -a <agent...>  # 或用 GitHub 上的最新代码(需 Git)
 ```
 
 `--dry-run` 预览, `--uninstall` 解除对应集成, 重新执行安装命令即为更新.
@@ -173,7 +174,7 @@ balance $362 | today $61.7 | 30d $566
 
 ### Vision(跨模型识图)
 
-让不支持图片的主模型借助多模态模型识图, 适用于读取报错截图、按设计稿还原 UI、定位测试反馈截图里的界面问题.
+让不支持图片的主模型借助多模态模型识图, 适用于读取报错截图, 按设计稿还原 UI, 定位测试反馈截图里的界面问题.
 
 #### 安装
 
@@ -213,23 +214,29 @@ npx -y @kairyou/agent-tools@latest vision -a claude codex opencode
 npx -y @kairyou/agent-tools@latest inspect-image <path|url> -q "导航栏的背景色和高度是多少"
 ```
 
-## 从 Git 运行
+## Extras
 
-如需直接使用仓库版本, 可将 npm 包名替换为 `github:kairyou/agent-tools` (本机需已安装 Git):
-
-```bash
-npx -y github:kairyou/agent-tools usage -a codex
-```
-
-贡献者可参考[仓库结构](docs/zh-CN/repository-structure.md).
+[extras](docs/zh-CN/extras.md) 收录场景较窄的内容.
 
 ## FAQ
 
-### 为什么全局安装提示 PromptScript 不支持?
+### 为什么安装 skills 时提示 PromptScript 不支持?
 
 `PromptScript does not support global skill installation` 表示 PromptScript Agent 不支持全局
 安装, 不影响其他 Agent, 可忽略. 参见 [`skills` issue #1352](https://github.com/vercel-labs/skills/issues/1352).
 
+### 无法访问 GitHub 时如何安装 skills?
+
+直接从 npm 包安装(`at-commit` 换成你要装的 skill):
+
+```bash
+cd "$(mktemp -d)" && tar -xf "$(npm pack @kairyou/agent-tools --silent)" && npx -y skills@latest add ./package --skill at-commit -g -y
+```
+
+### 想参与开发?
+
+参见[仓库结构](docs/zh-CN/repository-structure.md).
+
 ## 参考与致谢
 
 - `at-commit` 借鉴了 [OpenCommit](https://github.com/di-sukharev/opencommit) 和

package/config.default.jsonc

@@ -1,10 +1,6 @@
-// agent-tools config.
-// statusline.fields controls both visibility and order.
-// Remove a field to hide it; move a field to reorder it.
-// Available statusline fields: branch, model, fiveHour, week, context, directory.
-// providerUsage is used by the shared usage runtime.
 {
   "statusline": {
+    // Shown in this order; available: branch, model, fiveHour, week, context, directory
     "fields": ["branch", "model", "fiveHour", "week"],
     "separator": " | ",
     "symbols": {
@@ -17,7 +13,20 @@
     }
   },
   "providerUsage": {
-    "days": 30,
-    "debug": false
+    "days": 30,     // recent days of spend to count
+    "debug": false  // true: log probes to ~/.agent-tools/logs/usage-debug.log
   }
+
+  // Optional blocks below: uncomment every line of a block, including its leading comma.
+
+  // at-self-eval, at-daily-log: default projects (otherwise just the current repo)
+  // , "workProjects": [
+  //   "C:\\projects\\project-a",
+  //   "C:\\projects\\project-b"
+  // ]
+
+  // at-daily-log: default file when asked to record; at-self-eval reads it as fallback
+  // , "dailyLog": {
+  //   "output": "C:\\projects\\todo.md"
+  // }
 }

package/docs/en/extras.md

@@ -0,0 +1,63 @@
+# Extras
+
+Fully usable, just situational. Install each on demand.
+
+## at-self-eval
+
+Summarize git history, a work log you provide, or both into a self-evaluation
+for performance cycles (quarterly, semi-annual, promotion).
+
+```bash
+npx -y skills@latest add kairyou/agent-tools --skill at-self-eval -g -y
+```
+
+Usage:
+
+- `/at-self-eval` — the current Git user's work this quarter
+- `/at-self-eval [<author>] [--from yyyy-mm-dd] [--to yyyy-mm-dd]` — pick author and range
+- Plain language works too: say "summarize the first half of the year" instead of writing flags
+- `/at-self-eval summarize July for C:\projects\project-a and C:\projects\project-b` — only those projects
+- `/at-self-eval also include C:\projects\project-c` — add to the default scope
+- Paste a daily/weekly log, or give a file path, to add business context
+- `/at-self-eval summarize from C:\projects\todo.md only` — log only, Git is not read
+
+Review the generated result before use.
+
+## at-daily-log
+
+Summarize the day's Git activity across projects into a daily work log.
+
+```bash
+npx -y skills@latest add kairyou/agent-tools --skill at-daily-log -g -y
+```
+
+Usage:
+
+- `/at-daily-log` — print today's log
+- `/at-daily-log 2026-07-31` — pick the date
+- `/at-daily-log summarize today for C:\projects\project-a` — only that project
+- `/at-daily-log also include C:\projects\project-c` — add to the default scope
+- `/at-daily-log record the log` — record to the configured file; rerunning a day only refreshes the generated part
+- `/at-daily-log record it to C:\projects\todo.md` — record to that file
+- `/at-daily-log record every workday at 18:00` — guides you through an OS scheduled task
+
+Review the generated result before use.
+
+## Optional configuration
+
+All in `~/.agent-tools/config.jsonc`:
+
+```jsonc
+{
+  // at-self-eval, at-daily-log: default projects (otherwise just the current repo)
+  "workProjects": [
+    "C:\\projects\\project-a",
+    "C:\\projects\\project-b"
+    // "C:\\projects\\temporarily-disabled"
+  ],
+  // at-daily-log: default file when asked to record; at-self-eval also reads it as fallback
+  "dailyLog": {
+    "output": "C:\\projects\\todo.md"
+  }
+}
+```

package/docs/zh-CN/custom-gateway-routes.md

@@ -36,7 +36,7 @@ export async function run(context, { requestJson, agentConfig }) {
   const session = await login.json();
 
   // requestJson 会解析 JSON, 并在非 2xx 响应时抛错; 需要时可在这里传入
-  // authorization、cookie 等自定义认证 header.
+  // authorization, cookie 等自定义认证 header.
   const me = await requestJson(`${context.baseUrl}/api/user/self`, {
     headers: { authorization: `Bearer ${session?.data?.accessToken}` },
   });

package/docs/zh-CN/extras.md

@@ -0,0 +1,62 @@
+# Extras
+
+完整可用, 只是场景较窄. 按需单独安装.
+
+## at-self-eval
+
+把 git 历史或你提供的工作日志归纳成自评产出清单, 用于季度/半年/晋升等绩效周期.
+
+```bash
+npx -y skills@latest add kairyou/agent-tools --skill at-self-eval -g -y
+```
+
+用法:
+
+- `/at-self-eval` — 默认统计当前 Git 用户本季度的工作
+- `/at-self-eval [<作者>] [--from yyyy-mm-dd] [--to yyyy-mm-dd]` — 指定作者与日期范围
+- 在对话里直接说 `统计上半年的工作` / `统计1-3月的产出` 即可, 无需手写参数
+- `/at-self-eval 统计 C:\projects\project-a 和 C:\projects\project-b 的 7 月工作` — 只统计指定项目
+- `/at-self-eval 另外包含 C:\projects\project-c` — 在默认范围上追加
+- 可粘贴日报或周报文本, 也可提供文件路径补充业务背景
+- `/at-self-eval 只根据 C:\projects\todo.md 总结` — 只凭日志生成, 不读 Git
+
+生成的结果请人工核对.
+
+## at-daily-log
+
+汇总一天内各项目的 Git 活动, 生成工作日报.
+
+```bash
+npx -y skills@latest add kairyou/agent-tools --skill at-daily-log -g -y
+```
+
+用法:
+
+- `/at-daily-log` — 输出今天的日报
+- `/at-daily-log 2026-07-31` — 指定日期
+- `/at-daily-log 统计 C:\projects\project-a 今天的工作` — 只统计指定项目
+- `/at-daily-log 另外包含 C:\projects\project-c` — 在默认范围上追加
+- `/at-daily-log 记录日报` — 记录到配置的文件; 同一天重复执行只更新生成的部分
+- `/at-daily-log 记录到 C:\projects\todo.md` — 记录到指定文件
+- `/at-daily-log 每个工作日 18:00 自动记录` — 引导搭建系统定时任务
+
+生成的结果请人工核对.
+
+## 可选配置
+
+都在 `~/.agent-tools/config.jsonc`:
+
+```jsonc
+{
+  // at-self-eval, at-daily-log: 默认统计的项目(不配则只统计当前项目)
+  "workProjects": [
+    "C:\\projects\\project-a",
+    "C:\\projects\\project-b"
+    // "C:\\projects\\temporarily-disabled"
+  ],
+  // at-daily-log: 记录日报的默认文件; 未提供日志时 at-self-eval 也读它作补充
+  "dailyLog": {
+    "output": "C:\\projects\\todo.md"
+  }
+}
+```

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@kairyou/agent-tools",
-  "version": "0.10.1",
-  "description": "Reusable Agent Skills and installable integrations (statusline, provider usage, vision) for Codex, Claude Code, and opencode.",
+  "version": "0.11.0",
+  "description": "Reusable Agent Skills, plus integrations (statusline, provider usage, vision) that install into Codex, Claude Code, and opencode.",
   "license": "MIT",
   "repository": {
     "type": "git",

package/skills/workflow/at-daily-log/SKILL.md

@@ -0,0 +1,99 @@
+---
+name: at-daily-log
+description: "Summarize one day's Git activity into a concise daily work log. Uses the current repository, optional configured work projects, or paths named in conversation; configuration is never required."
+argument-hint: "[<date>]"
+---
+
+# Daily Work Log
+
+`resolve evidence -> group related commits -> draft or record`
+
+## Date and evidence scope
+
+Default to today. Accept plain-language dates or `YYYY-MM-DD`; query through the next
+day because Git's `--until` boundary is exclusive. State the resolved date.
+
+Read optional `workProjects` from `~/.agent-tools/config.jsonc`.
+
+- No project named: current Git repository plus configured projects.
+- Projects named directly: only those projects.
+- "Also include" / `另外包含`: add them to the default scope.
+
+Resolve paths to Git roots and deduplicate them. Report invalid paths; a non-Git current
+directory does not block other valid projects. Do not clone remote URLs without consent.
+
+For each repository, resolve the author with `git -C <root> config user.name`; never
+infer aliases or use the remote login as the author. Filter commits by comparing `%an`
+literally, not via an `--author` regex. Inspect all local branches, current
+HEAD, user-named branches, and configured upstreams. Unless the user requests local-only
+data, refresh only the configured upstream branches, grouped into one best-effort
+`git fetch --no-tags <remote> <branch...>` per involved remote. Fetch failure is
+non-fatal. Do not change the working tree or local branch history. Deduplicate commits
+by hash.
+
+## Build work items
+
+Collect non-merge commits for the resolved author and day. Use subjects and changed
+paths to turn related commits into concrete completed work items. Fold formatting,
+version bumps, and follow-up fixes into the outcome they supported; do not inflate one
+change into several deliverables.
+
+Use file counts and added/deleted lines only when they meaningfully support the work
+item. Generated files, lockfile churn, renames, and bulk formatting often make those
+numbers misleading. Metrics are evidence, never hours, difficulty, impact, or a
+productivity score.
+
+Current uncommitted changes may be shown separately as **in progress**, but never as
+proof of work on a past date. User-provided non-code work must remain clearly identified
+as user context.
+
+## Output
+
+Match the user's language and omit empty sections:
+
+```markdown
++ 2026-07-31
+  完成 2 项:
+  1. agent-tools: 完成多项目日报规则和 skill 落地.
+  2. vscode-plugin: 修复 Webview 刷新后状态丢失问题, 补充回归验证.
+
+  进行中 1 项:
+  - iunit-web: 推进导入流程重构, 已完成数据解析.
+
+  汇总: 3 个项目, 2 项完成, 1 项进行中.
+```
+
+Prefer outcomes over raw Git metrics. If no verified activity exists, say so rather
+than fabricate an entry.
+
+## Draft or record
+
+"Generate" / `生成日报` returns a draft. "Record" / `记录日报`, or an explicit
+write request, writes after previewing the entry. Resolve the destination in order:
+
+1. a path supplied in conversation;
+2. optional `dailyLog.output` in `~/.agent-tools/config.jsonc`;
+3. no destination: return the draft without writing.
+
+Read the destination before editing. Wrap each date's generated content in that date's
+own markers, `<!-- log:2026-07-31:start -->` / `<!-- log:2026-07-31:end -->`; the date
+line and every line outside the markers belong to the user. For an existing date, show
+the current block and the regenerated one before writing, then replace only that
+date's block. If the date exists without markers, append a marked block below the
+user's lines instead of editing them. Do not duplicate the date. On duplicate or
+unpaired markers, stop and propose the edit instead of writing. When Git shows no
+activity, leave the file unchanged and say so; the user can add a manual entry
+themselves.
+
+Scheduling is separate; set it up only when the user asks. Prefer the OS scheduler
+(Task Scheduler, cron, launchd) over agent-internal timers, which stop with the agent:
+schedule a headless run of the CLI this skill is executing in, invoking the skill with
+a recording request (in Claude Code, `claude -p "/at-daily-log record the log"`; other
+CLIs have their own headless form). Keep projects and output in config so the command
+stays stable. Before registering, confirm the schedule (suggest workdays),
+make sure the output file is resolvable, and run the exact command once; register only
+after that test run records correctly, and show how to remove the task. Headless auth
+differs from the interactive session, so a failed test run means stop instead of
+registering, and show the exact command, its error output, and the likely fix (log in
+for headless use, adjust the output path). An unattended run has nobody to confirm with, so it must follow the
+marker rules exactly and skip any file it cannot edit that way.

package/skills/workflow/at-self-eval/SKILL.md

@@ -0,0 +1,105 @@
+---
+name: at-self-eval
+description: "Summarize a contributor's Git history, a provided work log, or both into a concise, review-friendly self-evaluation for quarterly, semi-annual, or promotion cycles."
+argument-hint: "[<author>] [--from yyyy-mm-dd] [--to yyyy-mm-dd]"
+---
+
+# Self-Evaluation Summary
+
+`resolve evidence -> group by business line -> deduplicate -> <=8 outcomes -> review`
+
+## Author and date window
+
+Normalize flags or plain language such as `统计1-3月的产出` and `统计上半年的工作`
+into `(author, from, to)`. Explicit input wins; fill only missing values:
+
+| User gave | Resolve to |
+| --- | --- |
+| `from` and `to` | use both |
+| only `from` | `to` = today |
+| only `to` | `from` = Jan 1 of that year |
+| no range | current quarter |
+
+Parse half-years, calendar quarters, month ranges, and rolling periods. A spoken
+period without a year means the current year, except months still ahead of today,
+which mean the previous year. Query through `<to + 1 day>` because Git's `--until` is exclusive.
+Always exclude merge commits.
+
+Use an explicit author when supplied; otherwise resolve it independently per repository
+with `git -C <root> config user.name`. Never infer aliases or use the remote login as
+the author. Filter commits by comparing `%an` literally, not via an `--author` regex. If no commits match, report that and suggest listing known commit authors;
+do not try spelling or language variants. When results exist, mention once that the
+user can provide other author names if needed.
+
+## Project evidence
+
+Read optional `workProjects` from `~/.agent-tools/config.jsonc`.
+
+- No project named: current Git repository plus configured projects.
+- Projects named directly: only those projects.
+- "Also include" / `另外包含`: add them to the default scope.
+
+Resolve paths to Git roots and deduplicate them. Report invalid paths; a non-Git current
+directory does not block other valid projects. Do not clone remote URLs without consent.
+
+Inspect all local branches, current HEAD, user-named branches, and configured upstreams.
+Unless the user requests local-only data, refresh only the configured upstream
+branches, grouped into one best-effort `git fetch --no-tags <remote> <branch...>` per
+involved remote. Fetch failure is non-fatal. Do not change the working tree or local
+branch history. Deduplicate commits by hash.
+
+Begin the result with the resolved author/window and, for multiple repositories, a
+plain-language list of the repositories checked. Mention remote-update failures without
+Git jargon. Do not show commit counts or a generic source line.
+
+## Optional logs
+
+A pasted daily/weekly log or explicit file path provides business context. If neither is
+supplied, optionally read `dailyLog.output` from config. Explicit conversation input
+always wins; a missing log is non-fatal.
+
+Commit-backed items may be summarized directly. Keep log-only work separate for user
+confirmation; include it only after confirmation, using only dates stated in the log or
+by the user. Never turn undated log text into work inside the selected window.
+
+When the user asks for a summary from a log alone, skip project scanning; when the
+window has no Git evidence, fall back to the log. Either way the log becomes the
+primary source: say so in the result and skip per-item confirmation, as the whole
+draft is the user's own account.
+
+If the user supplies a remote repository URL, ask before cloning it to a temporary
+directory and remove it afterward unless asked to keep it. Confirm before scanning every
+remote branch.
+
+## Group the work
+
+Group related commits into business-line or module outcomes. Fold lint, formatting,
+version bumps, repeated syncs, and follow-up fixes into the outcome they supported.
+Drop isolated trivial housekeeping. Similar subjects alone are not duplicates; use the
+repository, changed paths, and intent, and surface uncertain attribution for review.
+
+## Write the summary
+
+Output a numbered list with no more than 8 top-level outcomes. Each item starts with the
+business line/module and states a concrete result with 2-4 key points or one concise
+sentence. Prefer high-impact delivery, architecture, and key fixes; condense or omit
+routine work. Do not distribute the quota evenly if that hides standout results.
+
+Mention a concentrated month only when it adds useful distribution context, never as a
+productivity score. Match the user's language. Every confirmed outcome must trace to a
+commit or a user-confirmed log item.
+
+```text
+> 作者: <author> · 窗口: 2026-01-01 ~ 2026-06-30 · (自动推断, 如不对请指正)
+> 项目: <repo1> · <repo2>
+
+1. <业务线A>: <结果>; <关键点>
+2. <业务线B>: <结果>
+```
+
+End with one reminder that this is a draft requiring review. Do not write it to a file
+unless asked. Add at most one extra hint, and only when evidence is likely incomplete:
+
+- unconfirmed log-only work exists;
+- terse commits lack business context and no log was available;
+- the conversation or log provides concrete evidence of unscanned projects/branches.