Skip to content
Back to skills

Init Project

ASecurity

Configure a starter project on setup request: detected project facts, only necessary user questions.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 7, 2026
ai-agentsnodegit

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned November 7, 2026

npx -y skills add ryanportfolio/winuse-mcp --skill init-project --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Init Project?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Init Project
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/ryanportfolio-init-project/badge)](https://www.skillsdirectory.com/skills/ryanportfolio-init-project)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: init-project
description: "Configure a starter project on setup request: detected project facts, only necessary user questions."
---

# Configure a starter

Run setup only when requested. FILL IN markers signal unknown facts, not authorization to edit the project. Inspect manifests, current project instructions, reference indexes and existing setup before asking questions. Reuse configured facts; never guess a deployment target or verification capability. Detect read-only and never switch the session's worktree to another branch; see [branch placement](#pitfall-branch-placement).

Ask only for consequential facts that cannot be detected. Draft or apply the requested narrow configuration and reference changes. Preserve the separate Codex and Claude runtime boundaries; shared architecture and commands can remain shared references. Avoid deleting template/plugin directories or adding remotes without checking their actual use and authorization.

For optional project profiles, full/minimal skill selection, prose defaults, template
cleanup, or starter-remote wiring, use [setup profiles](references/profiles.md). Preserve
useful capabilities and customizations; defaults are candidates, not automatic deletions.
Read ownership settings before changing discovery, including maintained Codex natives.

Give the project its own CI: run `node .claude/scripts/write-ci-workflow.mjs` to show the
detected stacks and the proposed `.github/workflows/ci.yml`, then write it with `--write`
after the user approves. If the file exists, show a diff and ask before `--force`.
Subfolder projects (monorepos) are not detected; adapt the file. See [project CI](references/ci.md).

Validate affected JSON, scripts, skill synchronization and links. Configuration changes can break behavior, so never recommend immediate merge merely because they are not app code. Report remaining unknowns. Setup does not imply dependency installation, commit, push, PR or deployment.

## Pitfall: branch placement

When the repo has an `origin` remote, setup commits land on a branch from freshly fetched `origin/main` (or the default branch); a local-only project with no remote yet keeps its current branch. Never switch the session's own worktree to another branch. If the app code lives only on an unmerged branch, read it without checking it out (`git show <ref>:<path>`, `git ls-tree -r <ref>`), then ask the user where the setup PR should land before writing anything. Switching the checked-out branch reloads skills and hooks from that branch, and a PR based on a feature branch carries its unmerged commits.

Files in this skill

  • SKILL.md2.5 KB
  • references/ci.md3.2 KB
  • references/profiles.md4.7 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in ai-agents →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.