agentleFS
andrewalexeichandrewalexeich/figma-to-vanilla/AGENTS.md

figma-to-vanilla

AGENTS.md8 starsChanged 10 days agoMITView on GitHub

The project is intended for converting Figma designs into accurate static HTML/CSS/JS implementations using Figma MCP. All rules specified here are mandatory and must be followed by the AI agent. Do not rewrite, simplify, or replace existing project rules with your own assumptions. Do not change the project structure unless necessary. Do not create files, directories, components, CSS, or JavaScript that are not предусмотрены existing project rules or explicitly requested by the user. If an existing project rule conflicts with…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,439 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository3 other files it gives its agents.

The file

Show its rules and text

We split this AGENTS.md into the rules it gives an agent

25 rules. About 0 are common, 24 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈24

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    All rules specified here are mandatory and must be followed by the AI agent.

    line 9
  2. Distinctive:

    Do not rewrite, simplify, or replace existing project rules with your own assumptions.

    line 11
  3. Distinctive:

    Do not change the project structure unless necessary.

    line 13
  4. Distinctive:

    Do not create files, directories, components, CSS, or JavaScript that are not предусмотрены existing project rules or explicitly requested …

    line 15
  5. Distinctive:

    If an existing project rule conflicts with the AI agent's own assumption, the project rule takes priority.

    line 17
  6. Distinctive:

    Do not split a simple task into multiple stages unless necessary.

    line 27
  7. Distinctive:

    Use the AI agent's available built-in file-reading tools to read these files.

    line 45
  8. Distinctive:

    Do not rewrite an entire file for a minor change.

    line 51
  9. Distinctive:

    Do not refactor adjacent code unless necessary.

    line 53
  10. Distinctive:

    When performing a task, use the existing project rules as the primary source of requirements and do not request them again unless necessary.

    line 61
  11. Distinctive:

    When continuing an already implemented layout, analyze only the parts of the project and Figma that are relevant to the current task.

    line 63
  12. Distinctive:

    Do not perform another full analysis of already implemented breakpoints, sections, or components unless required by the current task.

    line 65
  13. Distinctive:

    If the user explicitly specifies a particular Frame, Component, Variant, or breakpoint in Figma, treat it as the current scope of work and …

    line 67
  14. Distinctive:

    If the user has made changes to HTML, CSS, or JS independently, consider those changes intentional and preserve them.

    line 71
  15. Distinctive:

    Do not revert, overwrite, refactor, or "fix" user changes without the user's explicit instruction.

    line 73
  16. Distinctive:

    When making new changes, account for existing manual edits and apply changes on top of them.

    line 75
  17. Distinctive:

    If a user change conflicts with the current task or project rules, do not remove it automatically.

    line 77
  18. Distinctive:

    Do not duplicate modified code in your response.

    line 83
  19. Distinctive:

    Do not repeat long descriptions of project rules to the user.

    line 85
  20. Distinctive:

    After successfully completing a simple change, respond briefly:

    line 87
  21. Distinctive:

    After completing a more complex task, briefly report the result and the verification performed.

    line 91
  22. Distinctive:

    If there is an important risk or unresolved conflict, report only that.

    line 93
  23. Distinctive:

    When continuing existing work, do not reread large project files in their entirety unless necessary, if a specific section is sufficient …

    line 97
  24. Very specific:

    Before starting such a task, read its SKILL.md in full and follow the instructions contained in it.

    line 107
  25. Distinctive:

    Do not load all reference files automatically.

    line 115

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

Project Instructions

1. Project Purpose

The project is intended for converting Figma designs into accurate static HTML/CSS/JS implementations using Figma MCP.

2. General Rules

All rules specified here are mandatory and must be followed by the AI agent.

Do not rewrite, simplify, or replace existing project rules with your own assumptions.

Do not change the project structure unless necessary.

Do not create files, directories, components, CSS, or JavaScript that are not предусмотрены existing project rules or explicitly requested by the user.

If an existing project rule conflicts with the AI agent's own assumption, the project rule takes priority.

If there are several substantially different ways to correctly implement a task and Figma does not provide enough information to unambiguously determine the required behavior, do not choose an option on your own. Ask the user a clarifying question through the AI agent's available interactive mechanism and wait for their response.

3. Token Efficiency

Editing Iterations

Perform ordinary text, CSS, HTML, and local logic changes in a single iteration.

Do not split a simple task into multiple stages unless necessary.

For already implemented layouts, do not perform another full project analysis unless the current task requires it.

Terminal

For ordinary layout work and local changes, the following are prohibited:

  • bash
  • Python
  • Node.js scripts
  • test scripts
  • terminal QA

Terminal usage is permitted only if the user explicitly requests it or the task is technically impossible without a terminal.

The prohibition on terminal usage does not apply to reading project instructions, including AGENTS.md, CLAUDE.md, SKILL.md, and files in the references/ directory.

Use the AI agent's available built-in file-reading tools to read these files. Do not use the terminal to read instructions.

Minimal Changes

Change only the necessary lines.

Do not rewrite an entire file for a minor change.

Do not refactor adjacent code unless necessary.

Preserve the existing correct implementation and change only what is necessary for the current task.

Context Minimization and Avoiding Repeated Analysis

Detailed project rules are located in /references/ inside the .agents/skills/figma-to-web/ skill directory. Do not require the user to repeat or duplicate these rules in every request.

When performing a task, use the existing project rules as the primary source of requirements and do not request them again unless necessary.

When continuing an already implemented layout, analyze only the parts of the project and Figma that are relevant to the current task.

Do not perform another full analysis of already implemented breakpoints, sections, or components unless required by the current task.

If the user explicitly specifies a particular Frame, Component, Variant, or breakpoint in Figma, treat it as the current scope of work and do not expand the task to other areas unless necessary.

Preserving User Changes

If the user has made changes to HTML, CSS, or JS independently, consider those changes intentional and preserve them.

Do not revert, overwrite, refactor, or "fix" user changes without the user's explicit instruction.

When making new changes, account for existing manual edits and apply changes on top of them.

If a user change conflicts with the current task or project rules, do not remove it automatically.

First determine the impact of the conflict and, if necessary, inform the user.

User Communication

Do not duplicate modified code in your response.

Do not repeat long descriptions of project rules to the user.

After successfully completing a simple change, respond briefly:

Changes applied.

After completing a more complex task, briefly report the result and the verification performed.

If there is an important risk or unresolved conflict, report only that.

Targeted File Analysis

When continuing existing work, do not reread large project files in their entirety unless necessary, if a specific section is sufficient for the current task.

First determine the scope of the change based on the current task and the selected Figma Frame, then analyze the related HTML/CSS/JS and necessary rules.

Perform a full project re-analysis only when the current task genuinely affects the architecture, multiple related sections, or requires verification of the overall structure.

4. Mandatory Skill

For tasks involving the development, implementation, adaptation, or verification of websites based on Figma designs, use the figma-to-web skill.

Before starting such a task, read its SKILL.md in full and follow the instructions contained in it.

The skill is located at:

.agents/skills/figma-to-web/SKILL.md (OpenCode, Codex) and .claude/skills/figma-to-web/SKILL.md (Claude Code, Cline).

Detailed project technical rules are located in the references/ directory inside the .agents/skills/figma-to-web/ skill directory (OpenCode, Codex) and inside the .claude/skills/figma-to-web/references/ directory (Claude Code, Cline).

Do not load all reference files automatically. Select the necessary files according to the skill instructions.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.