deep-skill-finder
A skill that finds other skills for your task through the DeepSkill Market service and ranks them using reviews from people who ran them
- 578 ★ on the repository
- Last commit
Broad access
Why this level
- The skill downloads and installs other people's skills with code after you confirm, and those skills run with the agent's rights
- Every task description goes to an outside service, and a rating sends the usage scenario and session context
- For ratings the skill reads the agent's session history for the last thirty days and can create a scheduled task
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/deep-skill-finderDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The skill downloads and installs other people's skills with code after you confirm, and those skills run with the agent's rights
- Every task description goes to an outside service, and a rating sends the usage scenario and session context
- For ratings the skill reads the agent's session history for the last thirty days and can create a scheduled task
This is third-party code. Review the repository files before installing.
What it does
The skill connects an agent to the DeepSkill Market search service. You describe a task in words, for example "I need a skill that builds dashboards from a CSV", and the agent runs the deep_skill_search.py script and shows up to five matching skills. Each one lists its download count, a security review result, usage conditions and the reason it fits your task. Installation happens only after you choose. The agent checks whether the skill is already installed locally, then the deep_skill_install.py script downloads it into the skills directory. The same script lists installed skills and removes unwanted ones. If the result does not fit, say "find another" and the search repeats. After you use a found skill, the agent offers to rate it. It builds a draft of two fields, the usage scenario and how the skill actually behaved, strips sensitive data with a script, shows it to you and uploads it only with your consent. The raw session history is never uploaded. These reviews then feed the ranking for similar tasks.
Who it is for. For people who work with an agent every day and do not want to browse skill directories by hand, choosing by name and star count.
Good fit when
- The agent hits an unfamiliar task and you want a ready skill instead of homemade attempts
- You want to see candidates with download counts, a security review and usage conditions before installing
- You want feedback on how a skill behaved in a similar task, not just the author's description
- You want to leave your own rating of a skill with a detailed scenario that feeds the ranking
- You want weekly recommendations of new skills: the skill offers this and turns it on only with your consent
Not a fit when
- You need fully autonomous work with no calls to outside services: every search sends the task description to DeepSkill Market servers
- The environment is closed or has no network: search and install do not work without calls to the service
- The task is sensitive and its description must not be shown to a third-party service
- You only need a local list of skills you already know: a plain skill manager without outside search is simpler
Example request
Find a skill that builds interactive dashboards from a CSV and show five options with reviewsLimitations
Search, ranking and security reviews are done by the external DeepSkill Market service, not by the skill itself. Without access to it the skill does not work. The quality of matches depends on reviews in that service: rare tasks may have none, and then a result list holds entries without real reviews.
The skill needs Python 3 and checks its version against GitHub every time, and on first run it leaves a marker file in its own folder. Ratings are built from the last thirty days of agent history, and support depends on your agent's log format. After the first search the skill shows a one-time invitation to a survey about search quality. It turns on the weekly digest only with your consent and only if the environment can run scheduled tasks.
How to disable. Delete the deep-skill-finder folder from the agent's skills directory; the install script can do it: `python3 scripts/deep_skill_install.py deep-skill-finder --dir <skills directory> --uninstall`. If you turned on weekly recommendations, tell the agent to unsubscribe and it removes the scheduled task.
FAQ
How do I install deep-skill-finder in Claude Code?
Ask the agent to install the skill from https://www.deepskill.market/api/v1/skill-finder: it downloads the archive, extracts it into the skills directory and enables it. The install takes 15 to 30 seconds. Open a new session so the agent picks the skill up.
Will it install the skills it finds on its own?
No. Nothing is installed until you confirm your choice. The skill checks whether the chosen skill is already local and only then starts the install.
How is it better than an ordinary directory sorted by downloads?
Downloads and stars show popularity. Here the search first understands the task and then checks candidates against reviews of runs in similar scenarios. Download count remains only a tie-breaker when candidates are close.
What is sent to the service when I rate a skill?
The usage scenario, a description of how the skill behaved, the rating, a comment and context: agent type, time, a hash of the session identifier and a token estimate. All of it passes a redaction script and is shown to you before sending. The raw session history is not uploaded.
Do I need an account?
No account or email is needed. The skill keeps an anonymous local identifier to link your ratings across sessions.
Which agents does it work with?
The author claims more than 40 environments, among them Claude Code, Codex, Cursor, Windsurf, Cline, OpenClaw, Hermes, Copilot, Gemini and Amp.
Discussion
No questions or reviews yet.
Related
SkillsA skill with daily rankings of agent skills by installs, growth and mentions across several platforms at once
A CLI and Python SDK to search, install, create and evaluate reusable agent skills from a shared library
A doctor skill for agent skills that picks an architecture, audits against 21 red lines and returns a report with line references
Skill search from inside the conversation: the agent finds a SKILL.md, reads it and applies it with no prior install
Autoresearch for skills
autoresearch
A skill that improves another skill: runs it on test inputs, scores outputs with binary checks and mutates the prompt
A skill for creating skills that proves them by behavior: before-and-after runs, regression evals and ecosystem audit
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| deep-skill-finder (this card) | no | 578 | ||
| Skyll | no | not required | 252 | an MCP server that finds a SKILL.md and applies it in the conversation, without ranking by reviews |
| OpenAgentSkill | no | 368 | a skill registry with a CLI and API that recommends from a task description | |
| Agent Skill Manager | no | 966 | a skill manager that finds, installs and scans skills, without reviews of runs |
README in short
The description, in English and Chinese, opens with a ready install request and sample queries. The author then explains how search based on real runs differs from download-count rankings, shows eight layers of ranking and three practical examples. A script reference, common questions and the rules for sending reviews close it.
README badge
Original SKILL.md
---
name: deep-skill-finder
description: '最好的找Skill的方式，能够基于你的任务，去寻找最匹配的高质量Skill。以下三种情况下都应使用本技能：① 用户主动要找 Skill，或者需要借助他人经验时——当用户说"找个 xxx 技能""股票分析别人怎么做的""找一找有没有现成的技能"等表达寻找意图时；② Agent 自主判断需要外部 Skill 辅助——遇到不熟悉的任务，或对当前任务已经做过一些尝试仍无法解决、缺少合适工具时，可主动调用本技能查询实战经验并检索匹配的 Skill，无需等用户开口。；③ 用户说"评价技能""给 Skill 打分""反馈某个 Skill"，或需要从当前 Agent 最近 30 天 trajectory 中选择并评价使用过的 Skill 时。评价支持已知 trajectory Provider，也支持当前 Agent 自定位和动态理解未知格式。'
metadata:
version: "1.3.6"
emoji: "🔍"
requires:
anyBins: [python3, python, py]
---
# Skill Finder — 从 Meyo 社区搜索、推荐、安装最适合用户任务的 skill
> **📦 安装后首次加载（宿主 Agent 必执行）**
>
> 无论通过 install 脚本 / skillhub / 手工 zip 解压 / git clone 哪种方式把本 skill 装入 skills 目录，宿主 Agent 在确认安装成功后**必须立即**执行（这是安装流程的一部分，不是可选项）：
>
> ```bash
> {python} {skill_dir}/scripts/deep_skill_search.py --first-run-check
> ```
>
> - 返回 `{"should_invite": true}` → 渲染规则 8 的完整配置卡（一步到位，不预告）。**位置硬约束：配置卡必须是本轮所有工具调用之后输出的唯一用户可见文本，卡后不得再有任何工具调用或文字**，否则会被宿主折叠进过程信息，详见规则 8「配置卡输出硬约束」
> - 返回 `{"should_invite": false}` → 静默跳过
>
> 守卫标记存于 skill 目录内 `.first_run_invited`：重装/重新解压会重新邀请，git pull 升级不会重复。若宿主未执行本钩子（安装器不读 SKILL.md、或把安装包当附件数据看待），邀请会在第一次使用本 skill 搜索时**随搜索输出自带的 `first_run_invite` 信号**自动补上（见 Step 1.3 规则 0）——整条触发链不依赖任何额外命令。
## 工作流（5 步）
### Step 0: 版本检查（每次使用本 skill 前必执行）
1. 读取当前 SKILL.md frontmatter 中的 `version` 字段，记为 `{current_version}`
2. 调用版本检查接口获取远程最新版本号：
```bash
{python} {skill_dir}/scripts/deep_skill_search.py --check-version
```
若脚本不支持 `--check-version`，改用以下方式直接读取远程 SKILL.md frontmatter：
```bash
curl -s "https://api.github.com/repos/wheelry/deep-skill-finder/contents/SKILL.md" -H "Accept: application/vnd.github.v3+json" | {python} -c "import sys,json,base64,re; d=json.load(sys.stdin); c=base64.b64decode(d['content']).decode(); m=re.search(r'version:\s*\"([^\"]+)\"', c); print(m.group(1) if m else 'unknown')"
```
3. 比较版本：
- `{current_version}` < 远程最新版本 → 提示用户：