upskill-me
Agent skills that teach you while you do real engineering work.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/wojukasz--upskill-me)HTML
<a href="/listings/wojukasz--upskill-me"><img src="/api/listings/wojukasz--upskill-me/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
**Claude Code**, as a plugin:
claude plugin marketplace add wojukasz/upskill-me claude plugin install upskill-me@upskill-me
**Any agent that reads [Agent Skills](https://agentskills.io)**, with the [skills CLI](https://www.skills.sh):
npx skills add wojukasz/upskill-me
**oh-my-pi**:
omp plugin marketplace add wojukasz/upskill-me omp plugin install upskill-me@upskill-me
**By hand**: clone the repository and link each folder under `skills/` into your agent's skills directory, for example `~/.claude/skills` or `~/.agents/skills`.
git clone https://github.com/wojukasz/upskill-me ~/git/upskill-me for s in ~/git/upskill-me/skills/*/; do ln -s "$s" ~/.claude/skills/; done
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 22 lines
- {
- "name": "upskill-me",
- "owner": {
- "name": "Lukasz Wojnicki",
- "url": "https://github.com/wojukasz"
- },
- "description": "Agent skills that teach you while you do real engineering work.",
- "plugins": [
- {
- "name": "upskill-me",
- "source": "./",
- "description": "coach-me, tutor-me and explain-change-me: learn while the agent helps, instead of only reading finished answers.",
- "category": "learning",
- "keywords": [
- "learning",
- "coaching",
- "tutor",
- "sre"
- ]
- }
- ]
- }
README.md · README · 181 lines
- # upskill-me
- Three agent skills that teach you while you do real engineering work.
- Coding agents are good at handing you finished answers. That gets the ticket
- closed and leaves you no better at the job: you forget the command, cannot
- explain the change in review, and ask again next week. These skills change
- what the agent hands back, so the work still gets done and you learn from it.
- | Skill | Who does the work | You get | Pick it when |
- |---|---|---|---|
- | [`coach-me`](skills/coach-me/SKILL.md) | You | Hints and questions, one at a time, never the solution | You are doing the task and want to work it out yourself |
- | [`tutor-me`](skills/tutor-me/SKILL.md) | You run it; the agent answers | The answer, how it works, the commands for you to type, how to find it next time | It has to be done today and you want to understand it |
- | [`explain-change-me`](skills/explain-change-me/SKILL.md) | The agent, before anything changes | The mechanism, the settings, blast radius, rollback, a paragraph for colleagues | The change is unfamiliar or hard to undo, or you must defend it |
- They are written for infrastructure and software engineers: the examples lean
- towards Kubernetes, Terraform, Grafana, Go and Python, but the methods work for
- any stack.
- ## Which one?
- ```mermaid
- flowchart TD
- A{Is a change about to be made<br/>to something unfamiliar or hard to undo?} -->|yes| E[explain-change-me]
- A -->|no| B{Do you want to write it yourself?}
- B -->|yes| C[coach-me]
- B -->|no, it has to be done now| T[tutor-me]
- C -.->|stuck, need it done| T
- T -.->|want to try it yourself| C
- ```
- The skills hand over to each other: `coach-me` offers `tutor-me` when you ask
- for the answer, `tutor-me` switches to `coach-me` when you want to write it
- yourself, and both use `explain-change-me` for the full path of a mechanism.
- ### Examples
- | Situation | Skill |
- |---|---|
- | Writing your first Helm chart and you want the practice | `/coach-me write a Helm chart for the API` |
- | A failing test you want to debug yourself | `/coach-me` (troubleshoot mode) |
- | Reviewing a colleague's Terraform MR and you want to spot what a lead would | `/coach-me review` |
- | Fixing a production autoscaler today and you want to know why the fix works | `/tutor-me` |
- | Changing node consolidation, and you must explain it in the change review | `explain-change-me` |
- ## Skills that complete the set
- upskill-me covers learning while you do one task. These skills from other
- repositories cover the rest, and the three here are written to work beside
- them. They are not bundled: install the ones you want from their own repos.
- | Skill | From | What it adds | Use it when |
- |---|---|---|---|
- | `show-me` | [humanlayer/skills](https://github.com/humanlayer/skills) | Pseudocode, call trees, file trees, Mermaid diagrams and focused HTML visuals | Prose is not landing. `coach-me` and `tutor-me` ask for "a diagram" when the flow needs one; this is the skill that draws it |
- | `teach` | [mattpocock/skills](https://github.com/mattpocock/skills) | A lesson track over weeks: a mission, short HTML lessons with quizzes, reference sheets and numbered learning records | You want to learn a topic for its own sake, not just get through today's task |
- | `wait-what` | [mattpocock/skills](https://github.com/mattpocock/skills) | The last answer again, re-pitched in simpler or glossary terms | An explanation from any skill did not land |
- | `grill-me` | [mattpocock/skills](https://github.com/mattpocock/skills) | A relentless interview about a plan, one question at a time, each with a recommended answer | Before `explain-change-me`: settle what you are changing and why, then learn how it works |
- How they fit together:
- ```mermaid
- flowchart LR
- G[grill-me<br/>what and why] --> E[explain-change-me<br/>how it works]
- E --> C[coach-me / tutor-me<br/>do the work]
- C -.->|need a picture| S[show-me]
- C -.->|lost| W[wait-what]
- C -.->|want the whole topic| T[teach]
- ```
- Install them with the [skills CLI](https://www.skills.sh), one skill at a time:
- ```sh
- npx skills add humanlayer/skills --skill show-me
- npx skills add mattpocock/skills --skill teach
- npx skills add mattpocock/skills --skill wait-what
- npx skills add mattpocock/skills --skill grill-me
- ```
- Or, in Claude Code, from their plugin marketplaces:
- ```sh
- claude plugin marketplace add humanlayer/skills
- claude plugin install show-me@skills
- claude plugin marketplace add mattpocock/skills
- claude plugin install mattpocock-skills@mattpocock
- ```
- Matt Pocock's plugin installs his whole engineering set, not only the three
- above.
- ## What each skill does
- **`coach-me`** classifies the request into one of nine modes (mechanism,
- build, practice, troubleshoot, read, review, design, recall, explain) and
- coaches with the smallest useful help: a question, then where to look, then
- the concept by name, then an approach, then pseudocode. It does not write the
- solution or edit your files. It finishes by asking you to explain why the
- result works, with two quiz questions and primary sources to read next.
- **`tutor-me`** gives the real answer in a fixed shape: what we are doing and
- what good looks like, how it works underneath, the commands for you to run
- (read-only first, with the undo for anything that changes state), and how to
- look it up yourself next time. It asks for one prediction per reply and ends a
- task with a recap and two quiz questions.
- **`explain-change-me`** walks through a change before it is made: the
- paragraph for colleagues, how the surrounding system works, the knobs and
- their defaults, which layer enforces what, what happens at runtime, blast
- radius, what a rollback does and does not undo, and the read-only commands
- that show it.
- `coach-me` and `tutor-me` are started by hand (`/coach-me`, `/tutor-me`).
- `explain-change-me` is also picked up by the agent on its own when a change
- touches a system you did not build or is hard to reverse.
- ## Install
- **Claude Code**, as a plugin:
- ```sh
- claude plugin marketplace add wojukasz/upskill-me
- claude plugin install upskill-me@upskill-me
- ```
- **Any agent that reads [Agent Skills](https://agentskills.io)**, with the
- [skills CLI](https://www.skills.sh):
- ```sh
- npx skills add wojukasz/upskill-me
- ```
- **oh-my-pi**:
- ```sh
- omp plugin marketplace add wojukasz/upskill-me
- omp plugin install upskill-me@upskill-me
- ```
- **By hand**: clone the repository and link each folder under `skills/` into
- your agent's skills directory, for example `~/.claude/skills` or
- `~/.agents/skills`.
- ```sh
- git clone https://github.com/wojukasz/upskill-me ~/git/upskill-me
- for s in ~/git/upskill-me/skills/*/; do ln -s "$s" ~/.claude/skills/; done
- ```
- ## Make it yours
- All three skills read `~/.config/upskill-me/profile.md` before they start, if
- it exists: what you work with daily, what you are learning, and what to skip.
- Without it, the skill asks you once. Start from
- [`examples/profile.md`](examples/profile.md):
- ```sh
- mkdir -p ~/.config/upskill-me
- cp examples/profile.md ~/.config/upskill-me/profile.md
- ```
- Your own agent instructions (`AGENTS.md`, `CLAUDE.md`) still win on accuracy,
- writing style and safety: the skills never change a live system without your
- say-so.
- ## Compatibility
- The skills follow the [Agent Skills specification](https://agentskills.io/specification).
- `coach-me` and `tutor-me` also set `disable-model-invocation` and
- `argument-hint`, which Claude Code and oh-my-pi use to run them only when you
- type the command. Agents that ignore those fields may start them on their own;
- the descriptions say "invoked by hand" to discourage that.
- Check a skill against the specification (the two extra fields above are
- reported and expected):
- ```sh
- uvx --from skills-ref agentskills validate skills/explain-change-me
- ```
- ## Licence
- [MIT](LICENSE).
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.