repo-template

skill
Security Audit
Warn
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 6 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Baseline New Repository Template. .gitignore (multi-language). README.md (solo-focused). CLAUDE.md (with Copilot notes). CI workflow (.github/workflows/ci.yml) without review requirements. License. Basic structure.

README.md

repo-template

OpenSSF Best Practices
OpenSSF Scorecard
CI
License: MIT

A secure, agent-native GitHub repository template for Claude Code and Codex. The primary downstream workflow is: create the repository, point an agent that already understands the project at it, and let the agent normalize the template before materializing the real project.

[!IMPORTANT]
Agent inside a repository created from this template? If the repository is not vbonk/repo-template, do not treat this README, template examples, optional workflows, or generic documentation as project requirements. Read CLAUDE.md or AGENTS.md first and execute Phase 0. No command or setup interview is required.

The Golden Path

repo-template is optimized for a repository that has already been created from the GitHub template and is then entered cold by an engineering agent.

flowchart LR
    A[Create repo from template] --> B[Point Claude or Codex at repo]
    B --> C[Agent detects derived repository]
    C --> D[Phase 0: normalize template]
    D --> E{Project material already available?}
    E -->|Yes| F[Ingest current-session project]
    E -->|No| G[Leave clean intake-ready substrate]
    F --> H[Project-specific repository]
    G --> H

The agent should not make you explain the template, repeat project information it already has, or choose a stack merely because the repository is empty.

What Phase 0 does

Before application work, the first agent:

  1. verifies whether the repository is pristine or already contains user/project work;
  2. treats the template as scaffolding, not as application truth;
  3. classifies inherited artifacts as KEEP / ADAPT / REMOVE / DEFER;
  4. removes template identity, examples, false configuration, and irrelevant automation;
  5. preserves useful security and repository hygiene;
  6. uses project context already present in the current session instead of re-interviewing the user;
  7. either integrates that project immediately or leaves a truthful, minimal intake-ready substrate;
  8. verifies the resulting repository before committing.

See docs/PHASE-0.md for the canonical first-agent SOP.

Create a Repository

Use GitHub's Use this template button, or create one from the CLI if that is how you are already working:

gh repo create my-project --template vbonk/repo-template --private --clone

After creation, the recommended interaction is simply:

Organize and publish the project we have been working on to owner/repo.

A correctly configured Claude Code or Codex session should recognize the derived repository and handle Phase 0 automatically.

Agent Support

This template intentionally supports two agent surfaces deeply:

Agent Entry point Behavior
Claude Code CLAUDE.md Detects source-template vs. derived-repo mode; derived repos run Phase 0 automatically
Codex AGENTS.md Same bootstrap contract using the open AGENTS.md convention

Claude Code also receives .claude/ commands, skills, hooks, and agents. /project:bootstrap is an explicit fallback; normal derived-repo entry does not require the user to invoke it.

Design Principles

  • Agent-first, not wizard-first. Infer from repository state, current session context, and existing artifacts before asking questions.
  • Remove assumptions before adding implementation. No fake architecture, commands, environment variables, dependencies, or deployment targets.
  • Security survives cleanup. Template normalization must not casually remove secret protection, safe ignore rules, AI security boundaries, or other proven safeguards.
  • Truthful configuration only. A configuration that appears active but is not actually configured is worse than no configuration.
  • Minimal interruption. Ask only when a material decision is genuinely unknowable or unsafe to infer.
  • Repository as durable context. After intake, project-specific docs and agent instructions replace transient session knowledge.
  • Claude Code + Codex only. Depth over broad, shallow agent compatibility.

What Ships

The source template includes a broad security and governance toolkit so Phase 0 can retain the parts appropriate to the incoming project:

  • secret-aware .gitignore, .gitattributes, and .editorconfig;
  • Claude Code and Codex instruction surfaces;
  • pre-commit secret scanning and GitHub hardening scripts;
  • issue/PR governance and CODEOWNERS;
  • security documentation and prompt-injection defenses;
  • 19 workflows in the source template, including security, release, maintenance, and validation capabilities;
  • Claude commands, skills, hooks, and an example agent;
  • compliance and self-test tooling.

Not all of these belong in every derived project. Phase 0 decides what survives.

Source Template Development

If you are working on vbonk/repo-template itself, do not run destructive downstream normalization. Follow CLAUDE.md / AGENTS.md in source-template mode.

Useful validation commands:

bash scripts/test-template.sh --local-only
bash scripts/test-e2e.sh
bash scripts/audit-compliance.sh --local-only

Repository security hardening is audited with:

bash scripts/secure-repo.sh --audit

Documentation

Document Purpose
Phase 0 Canonical first-agent normalization and project-intake SOP
Getting Started Short agent-first usage guide
AI Security Agent threat model and security boundaries
Branch Protection Repository hardening and rulesets
Documentation Guide Documentation quality standard
Architecture Architecture of the template project itself
ADRs Material design decisions

The full index is in docs/README.md.

Example

repo-template-example demonstrates a fully initialized project derived from the template. It is an example, not a downstream requirement.

Security

Never commit secrets. Derived repositories should retain the template's security baseline unless Phase 0 can show that a control is irrelevant or has been replaced by an equivalent or stronger project-specific control.

See SECURITY.md and docs/AI-SECURITY.md.

License

repo-template itself is MIT licensed. A downstream project's license is a separate decision. Do not assume the application's licensing merely because the template uses MIT; preserve any notices required for template material that remains in the derived repository.

Reviews (0)

No results found