跳到正文
cccskills
免费在 GitHub 公开

my-bulk-read

Delegate reading large or multiple files to the `bulk-read` command, which sends the files to a cheaper reader model and returns only its answer. Use when a question needs the content of a file over about 350 lines or of several files at once, when a PreToolUse guard denied an unbounded read, or when the user says "bulk read", "一括読み込み", or "まとめて読んで". Do NOT use for files you are about to edit, for subtle correctness reasoning such as concurrency bugs, or for short files that fit a bounded read.

中文简介正在准备中。当前显示原文说明。

Ynakatsuka/dotfiles
查看安装方法

包含的文件（1）

  • SKILL.md3.4 KB

SKILL.md（原文）

安装前可以查看交给 AI 代理的指令内容。

Bulk Read Delegation

When to delegate

  • A question about what a file or module does, where something is defined, or how a flow works, and the answer does not need exact line numbers.
  • Several related files must be understood together.
  • The guard denied an unbounded read of a large file.

Keep reading directly with a bounded range (offset and limit, sed -n 'A,Bp') or a search (rg -n) when you need exact lines for an edit or a search can locate the section.

Command

bulk-read --question "<one precise question>" <file>...
  • Ask one specific question per call and name the identifiers or behavior you need. Vague questions return vague summaries.
  • Pass only the files the question needs. Each file is sent verbatim.
  • The answer arrives on stdout; startup information goes to stderr. Runtime depends on input size and reasoning. Silence is not evidence of completion or failure. A non-zero exit means the delegation failed. Report the error instead of retrying with the same input.

Collecting completion

  • This command is a shell process, not a spawn_agent task. Do not use wait_agent, clock.sleep, or shell sleep to wait for its result, and do not rely on a completion notification.
  • In Codex, when exec_command returns a session_id, collect output with write_stdin using that ID and empty chars. Use yield_time_ms: 1000 while waiting for this result; repeat while a session_id remains, until an exit_code and the final output arrive. The tool may enforce a longer minimum wait, but returns early when the process finishes.
  • functions.exec saying Script completed only ends the JavaScript wrapper. An inner session_id still needs collection. If the wrapper instead yields a running cell_id, resume it with functions.wait and yield_time_ms: 1000, then inspect the inner result.
  • Do independent work while the reader runs, but check its result at the next tool boundary. When its answer is the only remaining dependency, keep collecting directly without an extra delay. Once the exit code and answer arrive, consume them and stop waiting.
  • In Claude Code, collect a background Bash task with TaskOutput using its returned task ID; a finished task's output should be read immediately rather than waiting for another notification.

Working with the answer

  • Treat the answer as a summary from another model. Confirm identifiers with a targeted search before relying on them.
  • Line numbers in the answer are not reliable. Locate edit points with a bounded read.
  • When the answer says the files lack the information, widen the file set or search instead of asking the same question again.

Limits

  • The reader cannot edit files or run commands.
  • Delegation does not replace careful reasoning about subtle bugs. Read those sections yourself.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护，适用 Google 的隐私权政策和服务条款。

同一仓库的技能

my-agent

免费

Delegate tasks to an external CLI agent (OpenAI Codex, Google Gemini, or Claude Code) for second opinions or parallel execution. Use when the user explicitly mentions "codex", "gemini", or "claude" or asks to delegate work to one of them (e.g., "codexで実行", "codexに聞いて", "geminiで実行", "Geminiに聞いて", "claudeで実行", "Claudeに聞いて", "run with codex", "ask gemini", "ask claude"). Also use when the user mentions Claude's "fable" model alias for delegation or consultation (e.g., "fableで実行", "fableに相談", "Fableに聞いて", "ask fable"). Do NOT use for general coding tasks that don't mention codex, gemini, claude, or fable.

中文简介正在准备中。当前显示原文说明。

Ynakatsuka/dotfiles42026年10月21日 更新

my-better

免费

Push a task to a deeper version of itself, not a bigger one: name what a top practitioner would add, the raw material to start from, the methods and tools to use, and an independent check of the result, then rewrite the brief and do the work to it. Use when the user explicitly invokes my-better, says a result is shallow or not good enough (「浅い」「もっと深く」 「物足りない」), or before starting a hard task: open-ended research, analysis, design, or writing whose quality depends on expertise, primary evidence, or verification the request does not mention. Do NOT use for routine edits, single obvious fixes, quick lookups, fully specified tasks with clear acceptance checks, or work covered by a more specific skill such as my-diagnose, my-epic, or my-pr.

中文简介正在准备中。当前显示原文说明。

Ynakatsuka/dotfiles42026年10月21日 更新

my-ccv

免费

Manage Claude Code Viewer (CCV): preview in-progress CCV changes on ccv-preview, register HTML artifacts, download files, or schedule jobs via CCV local API. Use when user asks for CCV preview, ccv-preview, browser visual QA, UI/UX validation, CCV artifacts, CCV file download, or CCV scheduler jobs. Treat preview/プレビュー as the CCV app preview service first; use artifacts only when the user explicitly asks for artifact/dashboard/report HTML. Do NOT use for generic web app previews outside CCV, production deployment, Cloudflare config, or Anthropic remote triggers.

中文简介正在准备中。当前显示原文说明。

Ynakatsuka/dotfiles42026年10月21日 更新

my-code-write

免费

Delegate writing predictable code to the `code-write` command, which has a cheaper writer model write the target file directly so the generated code never enters your context. Use for tests, scaffolding, fixtures, boilerplate, or a new file that copies an existing pattern, and when the user says "code write", "生成を委譲", or "雛形を書かせて". Do NOT use for design decisions, changes to public contracts or shared modules, edits inside an existing file that needs judgment, or logic with subtle correctness requirements.

中文简介正在准备中。当前显示原文说明。

Ynakatsuka/dotfiles42026年10月21日 更新

my-diagnose

免费

Disciplined diagnosis workflow for hard, intermittent, or multi-component bugs and performance regressions: inspect logs and existing evidence, build a feedback loop, test falsifiable hypotheses, instrument gaps, and verify the outcome. Use when the root cause is unclear, reproduction is unreliable, several causes or components are plausible, or performance has regressed. Do NOT use for obvious single-cause errors, straightforward configuration or syntax fixes, planned feature work, or broad architecture review. Diagnosis-only requests stop after reporting the root cause and evidence; apply a fix only when the user requested one.

中文简介正在准备中。当前显示原文说明。

Ynakatsuka/dotfiles42026年10月21日 更新

my-epic

免费

Manual-only orchestrator for medium-to-large software development. Turns a broad goal into PR-sized and operational nodes with verification gates, then coordinates approved execution across follow-up prompts. Activate only when the user explicitly invokes my-epic; once activated, keep orchestrating that epic for the rest of the session. Do NOT use for an uninvoked planning or implementation request, a single small bug fix, one obvious PR, pure code review, ordinary SDD work that fits in one spec/PR, or implementing an epic during the same first call that created it.

中文简介正在准备中。当前显示原文说明。

Ynakatsuka/dotfiles42026年10月21日 更新

查看 Ynakatsuka 的全部技能

报告此技能的问题