skills.rest
gullitmiranda
gullitmirandaCOMMUNITY
Explore all skills by this author →
GitHub

persist-agent-constraints

Persist agent constraints into AGENTS.md or CLAUDE.md files.

2|Updated Oct 10, 2025
Search Tags:#automation#documentation#constraints#guardrails#persist#agent-instructions#dont-do
One-click install
npx skills add https://github.com/gullitmiranda/.cursor --skill persist-agent-constraints
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: persist-agent-constraints
Source: https://github.com/gullitmiranda/.cursor/tree/main/skills/persist-agent-constraints
Command: npx skills add https://github.com/gullitmiranda/.cursor --skill persist-agent-constraints

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Persists agent constraints (what not to do) in the project's canonical place for agent instructions. Use when the user states constraints, prohibitions, or things to avoid (e.g. não faça, evite, don't, never, avoid, não quero que, proibido).

Core Features & Use Cases

  • Persist constraints to the appropriate artifact in the project where agent instructions live (e.g., AGENTS.md, CLAUDE.md, .cursor/rules/*.mdc) or create one if none exists.
  • Contextually apply constraints across multi-project workspaces by targeting the relevant project root or repository, ensuring consistency.
  • If no suitable artifact exists, create a minimal AGENTS.md or CLAUDE.md at the project root and store constraints there.
  • Each constraint is stored as an individual bullet under a "Don't / Avoid" or "Constraints" section to keep them actionable and easy to audit.

Quick Start

Create or update the appropriate agent-instruction file with the new constraints.

Frequently Asked Questions about persist-agent-constraints

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I persist agent constraints in a project to keep AI aligned?▼

Agent constraints are persisted by writing stated prohibitions as individual bullets under a 'Don't / Avoid' or 'Constraints' section in canonical instruction files like AGENTS.md or CLAUDE.md, creating the file if none exists.

Where should I save AI guardrails and restrictions for multi-repo workspaces?▼

For multi-repo workspaces, AI guardrails and restrictions are saved into the most relevant artifact at the targeted project root, such as AGENTS.md, CLAUDE.md, or .cursor/rules/*.mdc, maintaining consistency across repositories.

What happens if there is no existing agent instruction file when I add constraints?▼

If no suitable artifact exists when adding constraints, a minimal AGENTS.md or CLAUDE.md file is automatically created at the project root to store the new AI restrictions and keep the workspace governed.

How do AI instructions work to prevent unwanted code generation actions?▼

AI instructions prevent unwanted actions by recording explicit prohibitions, such as 'never' or 'avoid', into project files that agents read before executing tasks, thereby enforcing boundaries on code generation and automation.

Can I use this to manage 'don't do' rules across multiple project repositories?▼

Yes, you can manage 'don't do' rules across multiple project repositories by contextually targeting the relevant project root, ensuring each constraint is applied to the correct repository's canonical agent instruction artifact.

Are there limitations to storing constraints in an AGENTS.md file?▼

Limitations include relying on the AI agent's ability to parse and adhere to the AGENTS.md file, meaning constraints are only effective if the agent reads and respects the 'Don't / Avoid' sections during task execution.

Related Skills

Explore More Software Engineering Skills→
legoutlegout
3

agent-md-refactor

Refactor monolithic agent instruction files into linked documentation hierarchies.

Community
Intermediate
AribSudiaAribSudia
2

arib-build

Decompose engineering goals into parallel specialist tasks and verify outputs against constraints.

Community
Advanced
dalagerdalager
1

github-multi-repo

Coordinate branches, pull requests, and version alignment across multiple GitHub repositories.

Community
Advanced
ybbms777ybbms777
1

setup

Automate review agent selection for compound-engineering projects using ce:review.

Community
Basic