- Home
- Skills
- Workflow Automation
- repo-seed
More screenshots
Works with the AI Agents you already use
Repo Seed
Empty folder in, real project out: plan, rulebook, memory, commands, reviewer agents and a working skeleton with CI, before any feature code.
Free
repo-seed
Example session with this skill installed
Set up a new project: a habit tracker web app with a FastAPI backend, just for me to start.
- Read your context and instructions
- Compiled the repo-seed
Asks the missing questions, proposes Next.js + FastAPI with layout, gateways and a 3-phase roadmap, waits for "ok", then writes CLAUDE.md, web/CLAUDE.md, api/CLAUDE.md, docs/PLAN.md, DESIGN.md, PROGRESS.md, INCIDENTS.md, docs/adr/, .claude/ (commands, agents, hook), both skeletons, Makefile, pre-commit and CI; lint and tests green; first commit; report with assumptions and your manual to-dos.
Connects securely to your tools. The creator never sees your data.
About this skill
Most AI-built projects go: prompt → accept → prompt → accept → why's it broken → start over. Nobody decided the architecture; it just happened.
repo-seed sets the project up before the first feature exists, and keeps you in charge of every decision.
- It asks what you're building (or reads your brief) — 9 short questions.
- It proposes the stack, folder layout, a gateway for every outside service, and a roadmap — then stops and waits for your OK.
- Once you approve, it writes the rules the agent must follow (
CLAUDE.md), the plan, session memory (PROGRESS.md,INCIDENTS.md, ADRs), 7 slash commands, 4 reviewer agents, a post-edit lint hook, and an empty-but-working skeleton with CI. - It loops until lint and tests pass, commits, and reports every assumption it made.
From then on every session starts with /start-session and ends with /end-session, so the agent always knows where the project is and why it's built that way.
Plain Markdown and one prompt. No install, no service. Open source (MIT): https://github.com/Aryan-yadav15/repo-seed
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Workflow Automation.
For new projects only (not existing repos). Writes no feature code. Setup is thorough rather than instant: it researches real package versions and loops until lint and tests pass. Opinionated: strict layering and a gateway per outside service; trim the rulebook if that's more than a weekend script needs.
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 buying a skill.
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 to use
No account required to browse
Trust & safety
Security scan
Security scan passed 1 day ago
- Free with an Agensi account
- Accounts
GitHub
- Software
GitAn empty project folder
Built for Claude Code (slash commands, agents, hook). Other agents can follow the generated CLAUDE.md and docs as plain instructions.
Permissions
Allowed Hosts
File Scopes
Clones the open-source kit into _kit/ in the project folder and deletes it at the end. Writes files only inside the project folder; runs git init and one commit.