一键导入
files
Identify files needed to accomplish a task
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Identify files needed to accomplish a task
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Make the current work resumable from a clean checkout — commit/push, record in-flight async work, flag what isn't durable. INVOKE ONLY when the user explicitly runs /handoff or asks for a handoff.
Launch and manage jobs on the HPC cluster (INVOKE ONLY when the user explicitly runs /hpc or asks to use this skill).
Analyze files for quality and consistency within the codebase
Deep analysis using parallel reasoning with iterative refinement
Deep verification that all changes meet requirements and preserve correctness
Read files in context
| name | files |
| disable-model-invocation | true |
| description | Identify files needed to accomplish a task |
| allowed-tools | Bash(find:*), Bash(grep:*), Bash(ls:*), Glob, Grep, Read |
| argument-hint | <task description> |
$ARGUMENTS
Identify the essential context files someone needs to read before starting this task — the files that rebuild the broader understanding an agent needs to avoid tunnel vision.
A fresh agent is about to work on this task with a clean context window. It will naturally discover specific implementation files as it works. Your job is different: identify the 3–6 files that provide the surrounding context — the big picture that prevents the agent from being over-focused on the immediate task and making mistakes from ignorance.
PLAN-foo.md)Output a short list with brief explanations:
path/to/file - why it matters for context
path/to/other - why it matters for context
Aim for ~6 files (3–6 typical; a few extra are allowed if truly important, e.g., for a fragmented knowledge base). After outputting the list, stop — do not read the files or start working.