agentleFS
Sign inStart free
Dragoon0xDragoon0x/argus/AGENTS.md

argus

AGENTS.md4 starsChanged 7 months agoMITView on GitHub

You are an AI coding agent working with the Argus design intelligence platform. When connected to the Argus MCP server, you have access to design intelligence tools that let you analyze and improve visual quality. These tools return structured, deterministic results — not opinions. Input: { url?: string, selector?: string, html?: string } Output: PageSnapshot with element positions, computed styles, accessibility data, and structural relationships. Use this when you need to understand what's visually happening on a page before making…

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

22 rules. About 0 are common, 12 are distinctive, and 10 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 10

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    When connected to the Argus MCP server, you have access to design intelligence tools that let you analyze and improve visual quality.

    line 7
  2. Distinctive:

    Use this when you need to understand what's visually happening on a page before making changes.

    line 17
  3. Distinctive:

    Use this to evaluate whether your generated UI meets quality standards before presenting it to the user.

    line 27
  4. Distinctive:

    Use this when the score is below threshold and you need concrete fixes.

    line 35
  5. Very specific:

    Use this to fix issues identified by design.suggest.

    line 43
  6. Distinctive:

    Use this after making changes to verify you improved things and didn't break anything.

    line 51
  7. Distinctive:

    Read the user's design taste profile.

    line 54
  8. Distinctive:

    Use this before generating UI to align with the user's established preferences.

    line 59
  9. Distinctive:

    Run a comprehensive design + accessibility audit.

    line 62
  10. Distinctive:

    Use this for thorough analysis when the user asks to evaluate or improve their design.

    line 67
  11. Very specific:

    Check Argus installation and configuration health.

    line 70
  12. Very specific:

    Call design.preferences to understand the user's taste

    line 78
  13. Distinctive:

    Use the profile data to inform your CSS choices

    line 79
  14. Very specific:

    After generating, call design.score to verify quality

    line 80
  15. Very specific:

    Call design.inspect on the target area

    line 83
  16. Very specific:

    Call design.score to identify weak categories

    line 84
  17. Very specific:

    Call design.suggest for specific fixes

    line 85
  18. Very specific:

    Call design.compare to verify improvement

    line 87
  19. Very specific:

    Call design.audit for full analysis

    line 90
  20. Distinctive:

    Ask which category to prioritize

    line 92
  21. Distinctive:

    All findings follow this structure:

    line 107
  22. Distinctive:

    When exporting preferences for system prompts:

    line 123

≈ “Common” is an estimate: at least 1 in 5 of the 36,368 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

AGENTS.md — Argus Agent Context

You are an AI coding agent working with the Argus design intelligence platform.

Your capabilities via Argus

When connected to the Argus MCP server, you have access to design intelligence tools that let you analyze and improve visual quality. These tools return structured, deterministic results — not opinions.

Available MCP tools

design.inspect

Analyze a page or component's visual structure.

Input: { url?: string, selector?: string, html?: string } Output: PageSnapshot with element positions, computed styles, accessibility data, and structural relationships.

Use this when you need to understand what's visually happening on a page before making changes.

design.score

Score design quality against established principles.

Input: { snapshot: PageSnapshot, categories?: string[], threshold?: number } Output: ScoreReport with per-rule findings, category scores, and an aggregate score (0–100).

Categories: spacing, typography, color, hierarchy, accessibility, consistency.

Use this to evaluate whether your generated UI meets quality standards before presenting it to the user.

design.suggest

Get specific, actionable improvement suggestions.

Input: { snapshot: PageSnapshot, maxSuggestions?: number, focus?: string } Output: Array of Suggestion objects with element selectors, CSS changes, and expected score improvement.

Use this when the score is below threshold and you need concrete fixes.

design.apply

Apply a design fix and verify the improvement.

Input: { selector: string, changes: CSSChanges, verify?: boolean } Output: { applied: boolean, before: Score, after: Score, delta: number }

Use this to fix issues identified by design.suggest. The verify flag re-scores after applying to confirm improvement.

design.compare

Compare two states of a page.

Input: { before: PageSnapshot, after: PageSnapshot } Output: DiffReport with structural changes, score deltas, and visual regression indicators.

Use this after making changes to verify you improved things and didn't break anything.

design.preferences

Read the user's design taste profile.

Input: { action: "read" | "export", format?: "xml" | "json" } Output: DesignProfile with learned preferences, weighted rules, and category biases.

Use this before generating UI to align with the user's established preferences.

design.audit

Run a comprehensive design + accessibility audit.

Input: { url?: string, selector?: string, wcag?: "AA" | "AAA" } Output: Full AuditReport with all category scores, accessibility violations, and prioritized fix list.

Use this for thorough analysis when the user asks to evaluate or improve their design.

design.doctor

Check Argus installation and configuration health.

Input: {} Output: Diagnostic report with version info, configuration status, and any detected issues.

How to use these tools effectively

Before generating UI

  1. Call design.preferences to understand the user's taste
  2. Use the profile data to inform your CSS choices
  3. After generating, call design.score to verify quality

When improving existing UI

  1. Call design.inspect on the target area
  2. Call design.score to identify weak categories
  3. Call design.suggest for specific fixes
  4. Apply fixes via code changes
  5. Call design.compare to verify improvement

When the user says "make it better"

  1. Call design.audit for full analysis
  2. Present the top 3 findings to the user
  3. Ask which category to prioritize
  4. Apply fixes, re-score, show before/after

Score interpretation

RangeQualityAction
90–100ExcellentShip it
75–89GoodMinor tweaks optional
60–74FairAddress warnings
40–59PoorSignificant issues need fixing
0–39CriticalFundamental problems

Structured output format

All findings follow this structure:

<design_finding>
  <rule>spacing-consistency</rule>
  <severity>warning</severity>
  <element>.card-grid > .card</element>
  <message>Inconsistent horizontal gaps: 16px, 24px, 12px found. Expected uniform value.</message>
  <measured>3 distinct gap values</measured>
  <expected>1 consistent gap value</expected>
  <suggestion>Set gap: 16px on .card-grid (most common value)</suggestion>
</design_finding>

Design preference XML export

When exporting preferences for system prompts:

<design_preferences source="argus" version="2026.4.12">
  <rule category="spacing" weight="0.9">ALWAYS use 8px grid for all spacing values</rule>
  <rule category="typography" weight="0.85">PREFER heading weight 600 over 700</rule>
  <rule category="color" weight="0.8">NEVER use pure black (#000) for text — use #1a1a1a minimum</rule>
  <rule category="hierarchy" weight="0.75">ALWAYS ensure primary CTA is visually dominant by minimum 1.5x size ratio</rule>
  <rule category="consistency" weight="0.7">AVOID more than 5 distinct font sizes per page</rule>
</design_preferences>

Paste this block into any system prompt, CLAUDE.md, .cursorrules, or agent configuration. The agent will respect these learned preferences in all generated output.

Download now (ZIP)

  • 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 28,693 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 repositoryOne other file it gives its agents.

Does it work?

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

Reviews can't be read right now.

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.

Check your setup, free