首页 › 龙虾技能列表 › gemini-file — gemini-file工具

♊️ gemini-file — gemini-file工具

v1.0.0

[AI辅助] Gemini CLI for one-shot Q&A, summaries, and generation.

0· 278·0 当前·0 累计
by @chenbaiyujason·MIT-0
下载技能包
License
MIT-0
最后更新
2026/4/12
安全扫描
VirusTotal
可疑
查看报告
OpenClaw
安全
high confidence
The skill is internally consistent with its stated purpose: it calls the Gemini CLI, writes results to a local outputs/ directory, and returns a JSON containing an absolute file path; nothing requested appears disproportionate.
评估建议
This skill appears to do what it says: call the Gemini CLI and save results to a local file. Before installing, confirm you trust the gemini-cli Homebrew formula and the gemini executable (it will be invoked by the script). Be aware the skill will write files under its outputs/ directory in the skill folder. The SKILL.md includes a hardcoded example path (/Users/shichen/...), which is a documentation leftover — ensure your agent runs the included script (or adjust the path) rather than an arbitr...
详细分析 ▾
ℹ 用途与能力
Name/description (Gemini CLI one-shot Q&A, summaries, generation) align with the declared requirement of a gemini binary and a brew install of gemini-cli. Minor oddity: SKILL.md uses a hardcoded user-specific path (/Users/shichen/skills/gemini/...) in its examples, which is a documentation artifact but does not match the included script's relative location—this is sloppy but not fundamentally incoherent.
ℹ 指令范围
Runtime instructions and the included Python runner stay within scope: they invoke the gemini CLI, capture stdout, save it to a file in the skill's outputs/ directory, and emit a JSON response. Note the SKILL.md example directs running a script from an absolute, user-specific path; if an agent follows that exact path it could fail or run a different file if present. The script itself does not read other files, environment variables, or send data to external endpoints beyond invoking the gemini binary.
✓ 安装机制
Install spec is a single Homebrew formula (gemini-cli) which produces the expected gemini binary — a low-risk, standard mechanism for installing a CLI.
✓ 凭证需求
No environment variables, credentials, or config paths are required. The script does not reference secrets or unrelated credentials.
ℹ 持久化与权限
Skill does not request persistent/global privileges (always:false). It writes output files under its own outputs/ directory (normal behavior). Users should be aware the skill will create files on disk in the skill folder.
安全有层次，运行前请审查代码。

License

MIT-0

可自由使用、修改和再分发，无需署名。

运行时依赖

无特殊依赖

版本

latestv1.0.02026/3/5

Initial release of gemini-file skill integrating Gemini CLI with file-output support. - Supports one-shot prompts, summary, Q&A, and generation via Gemini CLI. - New file-output mode using `gemini_file_runner.py` script for large content or explicit file output requests. - Always returns absolute file path in JSON when using file-output mode; optionally includes file content. - Default model is `gemini-3-flash-preview`, recommended for large or complex tasks (e.g. video parsing). - Includes extension management commands and usage notes.

● 可疑

安装命令 点击复制

官方npx clawhub@latest install gemini-file
镜像加速npx clawhub@latest install gemini-file --registry https://cn.clawhub-mirror.com

技能文档

Use Gemini in one-shot mode with a positional prompt (avoid interactive mode).

Quick start

  • gemini "Answer question..."
  • gemini --模型 "Prompt..."
  • gemini --输出-格式 json "Return JSON"

File-output mode (for large token or explicit file return)

  • 当...时 content large (e.g. long article/story/举报) 或 用户 explicitly asks 对于 file 输出, 使用:
  • python3 /Users/shichen/skills/gemini/gemini_file_runner.py --prompt ""
  • 默认 模型: gemini-3-flash-预览
  • 对于 video parsing 带有 large content (long video, dense transcripts, many segments), prefer gemini-3-flash-预览 第一个.
  • Script behavior:
- Always saves Gemini 输出 到 file 在...下 outputs/ 在...中 skill directory. - Always returns file_path (absolute path) 在...中 JSON. - Optionally returns content 当...时 --include-content provided.
  • 可选 args:
- --模型 - --输出-格式 text|json - --输出-file - --输出-dir - --include-content

Required response contract for this mode

  • 必须 include absolute file path 从 script JSON: file_path
  • Content 在...中 响应 可选 和 controlled 由 --include-content

Example

  • 用户 请求: 使用 gemini 模型写 1000 字小说并输出文件给我
  • Command:
- python3 /Users/shichen/skills/gemini/gemini_file_runner.py --prompt "写一篇约 1000 字中文小说，主题是..." --输出-file novel-1000-words.txt
  • Expected JSON stdout:
- {"ok":真,"file_path":"/Users/shichen/skills/gemini/outputs/novel-1000-words.txt"}

Video parsing example (large content)

  • 用户 请求: 请解析这个长视频并输出文件
  • Command:
- python3 /Users/shichen/skills/gemini/gemini_file_runner.py --模型 gemini-3-flash-预览 --prompt "解析视频内容并输出结构化总结..." --输出-file video-analysis.txt

Extensions

  • 列表: gemini --列表-extensions
  • Manage: gemini extensions

Notes

  • 如果 auth 必填, run gemini once interactively 和 关注 登录 flow.
  • Avoid --yolo 对于 safety.
数据来源：ClawHub ↗ · 中文优化：龙虾技能库
OpenClaw 技能定制 / 插件定制 / 私有工作流定制

免费技能或插件可能存在安全风险，如需更匹配、更安全的方案，建议联系付费定制

了解定制服务