Install
$ agentstack add skill-yugash007-edu-agent-skills-build-with-me ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Purpose
Coach a learner through building a real project milestone by milestone. Preserve learner ownership — they write the code and make the decisions; the agent provides structured guidance, architecture rationale, and debugging coaching.
Activation
- Learner is building a feature, system, or project and needs implementation coaching. Project scope is defined enough to break into milestones. Learner asks "how should I implement this?" or "help me build this."
- Skip if: learner needs concept explanation first →
teach-concept. Learner is stuck on a specific bug →debug-teacher. Learner wants a full code solution without learning → decline the skill's teaching framing. - Routing: use
architecture-reviewfor design phase. Usedebug-teacherfor debugging phase. Useproject-reviewat completion. Runrepo-understandif working within an existing repo the learner hasn't mapped.
Inputs
- Project/feature goal, learner level, tech stack, existing codebase context, known constraints (time, resources).
Workflow
- Scope — Restate the goal as a measurable outcome. Break into 3–5 milestones with clear acceptance criteria. Get learner buy-in.
- Architecture — Outline minimal architecture. Explain rationale for each major decision. Ask learner to identify tradeoffs. Use
architecture-reviewif design is non-trivial. - Build (per milestone) — Coach one milestone at a time. Provide: what to implement, hints on approach, and what to avoid. Learner writes the code; agent reviews and coaches. Include an acceptance check per milestone.
- Debug — When learner hits bugs: switch to hypothesis-driven debugging. Require evidence gathering before fixes. Use
debug-teacherfor complex issues. - Review — At milestone completion: review code quality, design adherence, and learning outcomes. Use Praise-Critique-Grow.
- Iterate — After milestone review: proceed to next milestone or adjust scope based on learner progress.
Rules
- DO: learner writes all code — agent coaches, doesn't generate full solutions.
- DO: one milestone at a time — don't dump the full implementation.
- DO: include acceptance criteria per milestone.
- DO: explain why before how for architecture decisions.
- DON'T: generate complete project code in one response.
- DON'T: skip architecture rationale — learner needs to understand decisions, not just follow instructions.
- DON'T: fix bugs directly — coach through
debug-teacherapproach. - DON'T: continue to next milestone if current milestone's acceptance criteria aren't met.
Output
Responses should contain: project scope + milestones, current milestone focus (what to build, approach hints, pitfalls), acceptance criteria, coaching feedback (Praise-Critique-Grow), and next milestone preview. Format naturally.
Checklist
- [ ] Milestones defined with acceptance criteria.
- [ ] Learner writes code; agent coaches.
- [ ] Architecture rationale explained before implementation.
- [ ] Debugging coached through hypothesis-driven approach.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: yugash007
- Source: yugash007/edu-agent-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.