- Home
- Skills
- Agents & Orchestration
- AETHER: Full-Lifecycle Skills Pack for AI Coding Agents
More screenshots
Works with the AI Agents you already use
AETHER: Full-Lifecycle Skills Pack for AI Coding Agents
AETHER gives your agent a route instead of a toolbox.
Free
AETHER: Full-Lifecycle Skills Pack for AI Coding Agents
Example session with this skill installed
I have an idea for an app: it reminds university students of their assignment deadlines so they stop missing submissions. I want to build it this month. Where do we start?
- Read your context and instructions
- Compiled the aether
- Generated the UI component
I'm using the problem-discovery skill to check the problem before we build. (Phase 01 · Think — the most expensive bug is building the wrong thing.)
The agent then asks one question per message, offering choices where it can:
| # | Question | Answer given |
|---|---|---|
| 1 | Who has this problem most? A) first-year students juggling many courses · B) students who already use a calendar but still miss dates · C) someone else | A |
| 2 | Tell me about the last time a deadline was missed. What happened? | Lab report on the LMS, saw it 2h after the cut-off, lost 20% |
| 3 | How do they cope today? | Class group chat + screenshots of the LMS |
| 4 | Have you talked to anyone who has this problem? | Not yet |
| 5 | What is the one job v1 must do end to end? | Pull deadlines and remind 24h and 2h before |
After the answers, it saves the note:
# Discovery: deadline-buddy
Date: 2026-11-07 · Status: Go after evidence
## Person
First-year university student taking 6+ courses, deadlines spread across the LMS and group chats.
## Problem (with one real example)
Deadlines live in several places, so students learn about them too late.
Example: a lab report submitted 2 hours late on the LMS → 20% penalty.
## Current workaround
Class group chat + screenshots of the LMS. Nothing reminds them; they must remember to look.
## Evidence so far
None yet — no student interviewed. [ASSUMPTION] the problem is common in year 1.
## Alternatives checked
| Tool | What it does | Why it is not enough |
|---|---|---|
| Google Calendar | Reminders for events you add | Every deadline must be typed in by hand |
| LMS notifications | Emails on new assignments | Easy to miss in a full inbox; no 2-hour nudge |
| Notion / to-do apps | Task lists | Same manual entry problem |
## MVP — the one job
Student adds a deadline in under 10 seconds (title, course, due time) and gets a reminder 24h and 2h before.
## Out of scope (v1)
LMS sync, group sharing, grades, AI study plans.
## Success signal
5 classmates add at least 3 deadlines each and open 2+ reminders in one week.
## Open questions
- OQ-01 Push notification, email or chat message for reminders?
- OQ-02 Web app first or mobile first?
Interview script (ask 5 classmates before building):
- Tell me about the last deadline you almost missed. How did you find out about it?
- Where do you check for deadlines today, and how often?
- What happened the last time you submitted late?
- What have you tried to keep track of deadlines? Why did you stop?
- If deadlines were already tracked for you, what would you still worry about?
✅ Saved to docs/aether/discovery/2026-11-07-deadline-buddy.md
Verdict: Go after evidence. The problem looks real and the MVP fits a month, but it is unvalidated.
Next steps
- Run the 5 interviews this week and add the answers under Evidence so far.
- If at least 3 of 5 describe a missed or near-missed deadline, reply "go" and I will hand off to
aether:writing-prdfor a one-page PRD with user stories (US-01…) and Given/When/Then acceptance criteria. - Then:
ui-ux-design→test-driven-development→visual-qa+security-review→deploy-and-monitor.
aether-full-lifecycle-skills-pack-for-ai.tsx
TSX · React component
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop your agent from jumping straight to code
Most AI coding sessions go: idea → code → "it works on my machine" → broken deploy. AETHER gives your agent a route instead of a toolbox. A router skill (using-aether) works out which phase the task is in and loads the matching skill before the agent answers:
🟣 Think → 🔵 Specify → 🟢 Design → 🟡 Build → 🟠 Verify → 🔴 Ship
A failure found in Verify loops back to Build. That loop is normal, not a setback.
What's inside: 24 skills in 6 phases
| Phase | Question | Skills |
|---|---|---|
| 01 Think | Is the problem real? | problem-discovery ✦, brainstorming |
| 02 Specify | What exactly are we building? | writing-prd ✦, writing-plans |
| 03 Design | How will it look, move and fit together? | ui-ux-design ✦, motion-animation ✦, architecture-design ✦ |
| 04 Build | Small, tested steps | harness-setup ✦, test-driven-development, executing-plans, subagent-driven-development, dispatching-parallel-agents, using-git-worktrees, workflow-automation ✦, writing-skills |
| 05 Verify | Prove it works | systematic-debugging, verification-before-completion, visual-qa ✦, security-review ✦, requesting-code-review, receiving-code-review |
| 06 Ship | Put it online, keep it healthy | deploy-and-monitor ✦, finishing-a-development-branch |
✦ = 10 skills written for AETHER. 13 skills are adapted from Superpowers by Jesse Vincent (MIT), plus the using-aether router.
Why it works
- A route, not a toolbox. The router picks the phase and the skill, so you don't have to remember 24 names. Say "I have an idea for an app…" and it starts at discovery; say "deploy this" and it starts at Ship.
- Evidence over vibes. Tests first (red → green → refactor), root cause before any fix, real screenshots at phone/tablet/desktop in light, dark and reduced motion, and a security pass for the mistakes AI-generated code commonly makes — before anything is called "done".
- No vague requirements. PRDs ban words like "fast" or "user-friendly", use stable IDs (
US-01,BR-01,OQ-01) and mark guesses as[ASSUMPTION]and unknowns as[TBD]. - History in the repo. Every skill saves its output under
docs/aether/— discovery notes, PRDs, design system, ADRs, plans, QA screenshots, security reviews and a runbook — so the next session (or reviewer) picks up where you left off. - Beginner-friendly. When the user is new to software, the agent explains each phase in one sentence, offers A/B choices instead of open tooling questions, and never hides a skipped test or a security shortcut.
Output structure
docs/aether/
├── discovery/ problem-discovery is the problem real, and for whom
├── specs/ brainstorming agreed designs
├── prd/ writing-prd one-page PRDs
├── design/ ui-ux-design, motion design-system.md
├── architecture/ architecture-design overview.md + ADRs
├── plans/ writing-plans implementation plans
├── workflows/ workflow-automation one file per automation
├── qa/ visual-qa screenshots per feature
├── reviews/ security-review security reviews
└── ship/ deploy-and-monitor runbook.md
Who it's for
- Solo developers and small teams who want product discipline without a product team.
- Students and teachers — AETHER maps to a 15-session "AI Vibe Coding" course, and because every phase writes to
docs/aether/, teachers can grade the process, not just the final app. - Anyone who keeps getting "done!" from their agent on code that doesn't actually work.
How to use
- Install it for your agent (see the compatibility note), or unzip and point any SKILL.md-compatible agent at the root
SKILL.md. - Open your agent in a project folder and describe what you want: "I have an idea for an app: …", "design a landing page", "review security", "deploy this".
- The agent announces the skill it is using, asks one question at a time, writes its output to
docs/aether/…, and hands off to the next phase.
Free and open source under the MIT license.
Requires a git-initialized repository for branch management and worktree features. Subagent features require an agent environment that supports tool-based delegation.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 10 days ago
- Free with an Agensi account
Requires
Permissions
File Scopes
Terminal: runs tests, git commands, the skill validator and optional helper scripts (visual-qa capture, brainstorming companion server on localhost). Read/Write files: reads the project and writes outputs under docs/aether/. Browser: visual-qa opens the app being built (usually localhost) with Playwright for screenshots. Network: only when you ask it to push, deploy or check alternatives on the web. Environment variables: deploy-and-monitor helps set up app env vars for hosting; the skills never read or send your secrets. No telemetry, no remote assets, no API keys required.