B
BoredHF/albertskills
Skills that make Claude think like an expert, not just act like one.
★ 2·更新 2026-11-15
在 GitHub 查看 ↗链接已于 2026-11-17 核查 · HTTP 200 · 打开的是厂商自己的官网
- ★ 数取值
- 2 ★ · BoredHF/albertskills · 取自 GitHub 搜索 API · (2026-11-19 20:00:48 核对)
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型:开源项目
- 收录与活跃
- 收录于 2026-11-16 · 最近见到 2026-11-20 · 共见到 5 次 · 被 2 个来源列出
- 评分构成
- 综合分 203 · 2 ★ · 2 个来源 · repo
常见问题(答案都来自我们的记录)
它适合谁 / 用来做什么?
我们把它归在这些标签下: Agent 技能。标签来自条目自身,不是我们强加的分类。
谁维护它、还在更新吗?
仓库
BoredHF/albertskills;我们收录于 2026-11-16;最近见到 2026-11-20。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。README 原文(英文,未翻译) · 在 GitHub 读完整 README →
# albertskills
以下摘录自该仓库的 README(片段)
Skills that make Claude think like an expert, not just act like one.
Most skill packs are task recipes: how to plan, how to debug, how to review. The eleven thinking skills here are different. Each one changes how Claude thinks on any kind of work (code, writing, design, research, plans): it critiques itself, competes against itself, remembers its mistakes, and notices when it is stuck. The pack also carries the seven workflow skills I use to run bigger jobs.
## Before / after
The request: "Write a Python function parse_duration(s) that turns strings like '1h30m' into seconds."
Without a skill, Claude writes this. It looks finished.
python
def parse_duration(s):
"""Turn a string like '1h30m' into seconds."""
total = 0
for value, unit in re.findall(r"(\d+)([dhms])", s):
total += int(value) UNITS[unit]
return total
With red-team, Claude attacks its own draft before you see it, and runs every break:
RED TEAM: parse_duration(s) returns the exact number of seconds for strings like '1h30m'.
| # | Severity | Break | Proof | Fix |
| 1 | Fatal | "1.5h" returns 18000 (read as "5h") | ran it | decimals rejected |
| 2 | Fatal | "-1h" returns 3600, sign silently dropped | ran it | rejected |
| 3 | Fatal | "1h30" returns 3600, the 30 is dropped | ran it | whole string must match |
| 4 | Serious | "" and "abc" return 0, same as a real zero | ran it | ValueError |
| 5 | Serious | "1h1h" returns 7200 | ran it | repeated unit rejected |
| 6 | Serious | "1H" returns 0 | ran it | units are case-insensitive |
Re-attack: the fixes rejected "1h 30m5s" and raised KeyError on "1H". Both fixed.
Open: "1m1h" (out of order) is still accepted as 3660.
Nine breaks that would otherwise have shipped, three of them fatal, plus two bugs in the fixes themselves, caught on the second pass. That is real output from testing the skill, trimmed to six of its nine rows.
## Install
Send this to your agent of choice:
Install the albertskills skill pack for me. Clone https://github.com/BoredHF/albertskills into a temporary folder, copy every folder inside its skills/ directory into ~/.claude/skills/ (create it if it is missing), then delete the temporary folder. Finish by listing the skills you installed and showing me one example of how to use one.
Want it in one project only? Add: "Install into .claude/skills/ in this project instead."
Or install it as a Claude Code plugin (2.1.275 or later):
/plugin install albertskills --marketplace BoredHF/albertskills
Or copy the files yourself. For every project:
bash
d=$(mktemp -d) && git clone --depth 1 https://github.com/BoredHF/albertskills "$d" && mkdir -p ~/.claude/skills && cp -r "$d"/skills/ ~/.claude/skills/ && rm -rf "$d"
For one project only, run this from the project root:
bash
d=$(mktemp -d) && git clone --depth 1 https://github.com/BoredHF/albertskills "$d" && mkdir -p .claude/skills && cp -r "$d"/skills/ .claude/skills/ && rm -rf "$d"
To update, send the prompt or run the command again. Claude picks a skill up when your request matches its trigger. You can also just name it: "red-team this", "run a tournament on the headline".
## Thinking skills
| Skill | What it does | Triggers when |
| tournament | Writes 3–5 genuinely different versions, scores them blind against a rubric built for the task, merges the best parts. | The output matters and the first idea is rarely the best: "give me your best", "try a few approaches". |
Most skill packs are task recipes: how to plan, how to debug, how to review. The eleven thinking skills here are different. Each one changes how Claude thinks on any kind of work (code, writing, design, research, plans): it critiques itself, competes against itself, remembers its mistakes, and notices when it is stuck. The pack also carries the seven workflow skills I use to run bigger jobs.
## Before / after
The request: "Write a Python function parse_duration(s) that turns strings like '1h30m' into seconds."
Without a skill, Claude writes this. It looks finished.
python
def parse_duration(s):
"""Turn a string like '1h30m' into seconds."""
total = 0
for value, unit in re.findall(r"(\d+)([dhms])", s):
total += int(value) UNITS[unit]
return total
With red-team, Claude attacks its own draft before you see it, and runs every break:
RED TEAM: parse_duration(s) returns the exact number of seconds for strings like '1h30m'.
| # | Severity | Break | Proof | Fix |
| 1 | Fatal | "1.5h" returns 18000 (read as "5h") | ran it | decimals rejected |
| 2 | Fatal | "-1h" returns 3600, sign silently dropped | ran it | rejected |
| 3 | Fatal | "1h30" returns 3600, the 30 is dropped | ran it | whole string must match |
| 4 | Serious | "" and "abc" return 0, same as a real zero | ran it | ValueError |
| 5 | Serious | "1h1h" returns 7200 | ran it | repeated unit rejected |
| 6 | Serious | "1H" returns 0 | ran it | units are case-insensitive |
Re-attack: the fixes rejected "1h 30m5s" and raised KeyError on "1H". Both fixed.
Open: "1m1h" (out of order) is still accepted as 3660.
Nine breaks that would otherwise have shipped, three of them fatal, plus two bugs in the fixes themselves, caught on the second pass. That is real output from testing the skill, trimmed to six of its nine rows.
## Install
Send this to your agent of choice:
Install the albertskills skill pack for me. Clone https://github.com/BoredHF/albertskills into a temporary folder, copy every folder inside its skills/ directory into ~/.claude/skills/ (create it if it is missing), then delete the temporary folder. Finish by listing the skills you installed and showing me one example of how to use one.
Want it in one project only? Add: "Install into .claude/skills/ in this project instead."
Or install it as a Claude Code plugin (2.1.275 or later):
/plugin install albertskills --marketplace BoredHF/albertskills
Or copy the files yourself. For every project:
bash
d=$(mktemp -d) && git clone --depth 1 https://github.com/BoredHF/albertskills "$d" && mkdir -p ~/.claude/skills && cp -r "$d"/skills/ ~/.claude/skills/ && rm -rf "$d"
For one project only, run this from the project root:
bash
d=$(mktemp -d) && git clone --depth 1 https://github.com/BoredHF/albertskills "$d" && mkdir -p .claude/skills && cp -r "$d"/skills/ .claude/skills/ && rm -rf "$d"
To update, send the prompt or run the command again. Claude picks a skill up when your request matches its trigger. You can also just name it: "red-team this", "run a tournament on the headline".
## Thinking skills
| Skill | What it does | Triggers when |
| tournament | Writes 3–5 genuinely different versions, scores them blind against a rubric built for the task, merges the best parts. | The output matters and the first idea is rarely the best: "give me your best", "try a few approaches". |
收录徽章(可选)
想把收录标记放进 README 就复制这行,可选,不影响收录。
[](https://anclum.com/zh/s/boredhf-albertskills/)
- 来源
- github · https://github.com/BoredHF/albertskills
- 按领域浏览
- Agent 技能
- Traction
- 2 ★
- 收录来源
- 2 个独立来源
- 收录于
- 今天
更多条目
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。