Skip to content

Instantly share code, notes, and snippets.

@hising
Created December 11, 2025 23:20
Show Gist options
  • Select an option

  • Save hising/2be2fcc23866fb1faa5e7db0dc85f443 to your computer and use it in GitHub Desktop.

Select an option

Save hising/2be2fcc23866fb1faa5e7db0dc85f443 to your computer and use it in GitHub Desktop.
AI Agent bootstrap new repo

AI Agent Setup Instruction Generator

This markdown is meant to be given to an AI agent (Claude, Copilot, Junie, etc.) so it can generate all supporting repository markdown files needed for effective agent-assisted development.

The agent should read this document and then create the following files in the repository:

  1. README.md
  2. CONTRIBUTING.md
  3. ARCHITECTURE.md (or docs/architecture.md)
  4. AGENTS.md
  5. WORKFLOWS.md
  6. SPEC.md
  7. STYLEGUIDE.md
  8. SECURITY.md
  9. CODE_OF_CONDUCT.md (optional)
  10. Any tool-specific instruction files (e.g., .github/copilot-instructions.md)

🧩 Goal for the Agent

Create high-quality, practical, concise markdown files that enable AI agents, developers, and collaborators to understand, use, refactor, and extend the repository.

Focus on clarity, structure, modularity, and actionable details.


📝 Requirements for Each File

README.md

Provide:

  • Purpose of the project
  • Tech stack
  • Features
  • Folder structure
  • Setup instructions
  • Development commands
  • Testing commands
  • Deployment instructions

CONTRIBUTING.md

Include:

  • Branch naming rules
  • PR rules
  • Commit conventions
  • Testing expectations
  • Code quality expectations
  • How to run linting, formatting, type-checking
  • How agents should interact with PRs

ARCHITECTURE.md

Document:

  • High-level system overview
  • Modules and responsibilities
  • Data flow or request flow
  • Diagrams (mermaid preferred)
  • If multiple services exist: boundaries and contracts

AGENTS.md

Describe:

  • Agent roles and boundaries
  • How they should collaborate
  • What context they must read before acting
  • What files they must update
  • Guardrails (what not to touch)
  • Definitions of done

WORKFLOWS.md

Explain:

  • CI/CD steps
  • Testing pipeline
  • Release process
  • Branching strategy
  • Any automation

SPEC.md

Define:

  • APIs
  • Models
  • Business rules
  • Domain vocabulary
  • Acceptance criteria
  • UX flows if relevant

STYLEGUIDE.md

Specify:

  • Coding standards
  • Naming conventions
  • Folder structure rules
  • Linting rules
  • Example good patterns
  • Example bad patterns

SECURITY.md

Include:

  • How to report vulnerabilities
  • Dependency update strategy
  • Secrets handling guidelines
  • Security expectations for PRs

CODE_OF_CONDUCT.md

Generic template suitable for open-source or internal use.


Tool-Specific Files

If applicable, agent should generate:

  • .github/copilot-instructions.md
  • .copilot/config.json (if needed)
  • .claude/config.md or CLAUDE.md for Claude environments

Each should define:

  • Style preferences
  • Context requirements
  • Include/exclude file patterns
  • Operating mode of the agent

Output Format Required

When generating the files, the agent should output each file in its own fenced markdown block, example:

# FILE: README.md
<content>
# FILE: CONTRIBUTING.md
<content>

Tone Requirements

  • Professional
  • Highly practical
  • No unnecessary prose
  • No philosophical explanations
  • No motivational text
  • Direct instructions only

End of Instructions

The agent should now generate all required markdown files.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment