ornament-style-mono

Design monochrome ornamental patterns grounded in Alexander Speltz's classical ornament taxonomy. Covers historical period selection, motif structural analysis, prompt construction for line art and silhouette rendering, and AI-assisted image generation via Z-Image. Use when creating decorative borders, medallions, or friezes in a single color, exploring historical ornament styles through generative AI, producing line art or pen-and-ink renderings of classical motifs, or generating reference imagery for design or educational materials.

9 stars

Best use case

ornament-style-mono is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Design monochrome ornamental patterns grounded in Alexander Speltz's classical ornament taxonomy. Covers historical period selection, motif structural analysis, prompt construction for line art and silhouette rendering, and AI-assisted image generation via Z-Image. Use when creating decorative borders, medallions, or friezes in a single color, exploring historical ornament styles through generative AI, producing line art or pen-and-ink renderings of classical motifs, or generating reference imagery for design or educational materials.

Teams using ornament-style-mono should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/ornament-style-mono/SKILL.md --create-dirs "https://raw.githubusercontent.com/pjt222/agent-almanac/main/i18n/caveman-lite/skills/ornament-style-mono/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/ornament-style-mono/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How ornament-style-mono Compares

Feature / Agentornament-style-monoStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Design monochrome ornamental patterns grounded in Alexander Speltz's classical ornament taxonomy. Covers historical period selection, motif structural analysis, prompt construction for line art and silhouette rendering, and AI-assisted image generation via Z-Image. Use when creating decorative borders, medallions, or friezes in a single color, exploring historical ornament styles through generative AI, producing line art or pen-and-ink renderings of classical motifs, or generating reference imagery for design or educational materials.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

# Ornament Style — Monochrome

Design monochrome ornamental patterns by combining art historical knowledge of classical ornament with AI-assisted image generation. Every design is rooted in a specific historical period and motif tradition from Alexander Speltz's *The Styles of Ornament* (1904).

## When to Use

- Creating decorative borders, medallions, friezes, or panels in a single color
- Exploring historical ornament styles through generative AI
- Producing line art, silhouette, woodcut, or pen-and-ink renderings of classical motifs
- Generating reference imagery for design, illustration, or educational materials
- Studying the structural grammar of ornamental traditions across cultures and periods

## Inputs

- **Required**: Desired historical period or style (or "surprise me" for random selection)
- **Required**: Application context (border, medallion, frieze, panel, tile, standalone motif)
- **Optional**: Specific motif preference (acanthus, palmette, meander, arabesque, etc.)
- **Optional**: Rendering style preference (line art, silhouette, woodcut, pen-and-ink, engraving)
- **Optional**: Target resolution and aspect ratio
- **Optional**: Seed value for reproducible generation

## Procedure

### Step 1: Select Historical Period

Choose a period from the classical ornament taxonomy. Each period has characteristic motifs and structural principles.

```
Historical Ornament Periods:
┌───────────────────┬─────────────────┬──────────────────────────────────────────┬──────────────────────┐
│ Period            │ Date Range      │ Key Motifs                               │ Mono Suitability     │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Egyptian          │ 3100–332 BCE    │ Lotus, papyrus, scarab, winged disk,     │ Excellent — bold     │
│                   │                 │ uraeus, ankh                             │ geometric forms      │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Greek             │ 800–31 BCE      │ Meander/Greek key, palmette, anthemion,  │ Excellent — high     │
│                   │                 │ acanthus, guilloche, egg-and-dart        │ contrast geometry    │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Roman             │ 509 BCE–476 CE  │ Acanthus scroll, rosette, grotesque,     │ Very good — dense    │
│                   │                 │ candelabra, rinceau, trophy              │ carved relief style  │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Byzantine         │ 330–1453 CE     │ Interlace, vine scroll, cross forms,     │ Good — flat          │
│                   │                 │ basket weave, peacock, chi-rho           │ silhouette style     │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Islamic           │ 7th–17th c.     │ Arabesque, geometric star, muqarnas,     │ Excellent — pure     │
│                   │                 │ tessellation, knotwork, calligraphic     │ geometric abstraction│
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Romanesque        │ 1000–1200 CE    │ Interlace, beast chains, chevron,        │ Very good — heavy    │
│                   │                 │ billet, zigzag, inhabited scroll         │ carved stone quality │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Gothic            │ 1150–1500 CE    │ Trefoil, quatrefoil, crocket,           │ Very good — tracery  │
│                   │                 │ finial, tracery, naturalistic leaf       │ and window patterns  │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Renaissance       │ 1400–1600 CE    │ Grotesque, candelabra, putto,           │ Good — engraving     │
│                   │                 │ medallion, festoon, cartouche           │ and woodcut styles   │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Baroque/Rococo    │ 1600–1780 CE    │ C-scroll, S-scroll, shell, asymmetric   │ Moderate — complex   │
│                   │                 │ cartouche, garland, ribbon              │ forms benefit from   │
│                   │                 │                                          │ color for depth      │
├───────────────────┼─────────────────┼──────────────────────────────────────────┼──────────────────────┤
│ Art Nouveau       │ 1890–1910 CE    │ Whiplash curve, organic line, lily,     │ Excellent — defined  │
│                   │                 │ dragonfly, femme-fleur, sinuous vine    │ by line quality      │
└───────────────────┴─────────────────┴──────────────────────────────────────────┴──────────────────────┘
```

1. If the user specified a period, confirm and note its characteristic motifs
2. If "surprise me," select randomly — weight toward periods with "Excellent" mono suitability
3. Note 2-3 primary motifs associated with the period for use in prompt construction

**Got:** A clearly identified period with 2-3 candidate motifs and understanding of why the period's ornament works well (or presents challenges) in monochrome.

**If fail:** If the user requests a period not in the table (e.g., Celtic, Aztec, Art Deco), research its ornamental vocabulary using WebSearch or WebFetch and construct an equivalent entry with motif list and mono suitability assessment before proceeding.

### Step 2: Analyze Motif Structure

Understand the structural grammar of the chosen motif before constructing the prompt.

1. Identify the **symmetry type**:
   - Bilateral (mirror across one axis — most organic motifs)
   - Radial (rotational — rosettes, medallions, star patterns)
   - Translational (repeating unit — friezes, borders, tessellations)
   - Point (central focus radiating outward — compass roses, mandalas)

2. Identify the **geometric scaffold**:
   - Circle-based (rosettes, medallions, roundels)
   - Rectangle-based (panels, metopes, cartouches)
   - Triangle-based (pediment fills, spandrels)
   - Band-based (friezes, borders, running ornament)

3. Identify the **fill pattern**:
   - Solid (silhouette, no internal detail)
   - Line-filled (hatching, cross-hatching, parallel lines)
   - Open (outline only, negative space dominant)
   - Mixed (outline with selective internal detail)

4. Identify the **edge treatment**:
   - Clean boundary (contained within a frame)
   - Organic bleed (motif extends beyond or dissolves at edges)
   - Interlocking (connects to adjacent units — for repeating patterns)

**Got:** A structural description like "bilateral symmetry, band-based scaffold, line-filled, interlocking edges" that will inform the prompt.

**If fail:** If the motif structure is unclear, look up visual references using WebSearch for "[period] [motif] ornament" and analyze the first few results. Speltz's original plates are public domain and widely available online.

### Step 3: Construct Monochrome Prompt

Build the text prompt for Z-Image generation using the period, motif, and structural analysis.

**Prompt Template:**
```
[Rendering style] of [motif name] ornament in the [period] style,
[composition type], monochrome, black and white,
[structural details from Step 2],
[application context], [additional qualifiers]
```

**Rendering Style Options:**
- `detailed line art` — clean vector-like lines, no fills
- `black silhouette` — solid black forms on white ground
- `woodcut print` — bold carved lines with wood grain texture
- `pen-and-ink illustration` — fine lines with hatching for depth
- `copperplate engraving` — precise parallel lines creating tonal gradation
- `stencil design` — connected negative space, no floating islands

**Composition Qualifiers:**
- `symmetrical`, `centered`, `repeating pattern`, `border design`
- `isolated motif on white background`, `continuous frieze`
- `within a circular frame`, `filling a rectangular panel`

**Monochrome Constraint (always include):**
- `monochrome, black and white, no color, no shading` (for pure line art)
- `monochrome, black and white, high contrast` (for silhouette)
- `monochrome, black and white, fine hatching for depth` (for engraving style)

**Example Prompts:**
- `detailed line art of Greek meander border pattern, continuous frieze, monochrome, black and white, geometric precision, repeating unit, classical antiquity style`
- `black silhouette of Egyptian lotus and papyrus ornament, symmetrical panel design, monochrome, black and white, high contrast, temple decoration style`
- `pen-and-ink illustration of Art Nouveau whiplash curve with lily motif, vertical panel, monochrome, black and white, sinuous organic lines, Alphonse Mucha influence`

**Got:** A prompt of 20-40 words that specifies rendering style, motif, period, composition, and monochrome constraint.

**If fail:** If the prompt is too vague, add structural specifics from Step 2. If too complex (over 50 words), simplify by removing adjectives and keeping only the structural essentials. Z-Image responds best to clear, specific prompts — avoid abstract or conceptual language.

### Step 4: Configure Generation Parameters

Select resolution and generation parameters appropriate to the application context.

```
Resolution by Application:
┌────────────────────┬─────────────────────┬────────────────────────────────┐
│ Application        │ Recommended         │ Rationale                      │
├────────────────────┼─────────────────────┼────────────────────────────────┤
│ Medallion / Roundel│ 1024x1024 (1:1)     │ Radial symmetry needs square   │
│ Tile / Repeat Unit │ 1024x1024 (1:1)     │ Square for seamless tiling     │
│ Horizontal Frieze  │ 1280x720 (16:9)     │ Wide format for running border │
│ Vertical Panel     │ 720x1280 (9:16)     │ Portrait format for columns    │
│ Wide Border        │ 1344x576 (21:9)     │ Ultrawide for architectural    │
│ General / Flexible │ 1152x896 (9:7)      │ Balanced landscape format      │
│ Large Detail       │ 1536x1536 (1:1)     │ Higher res for fine line work  │
└────────────────────┴─────────────────────┴────────────────────────────────┘
```

1. Select resolution based on application context
2. Set `steps` to 8 (default) for initial generation; increase to 10-12 for fine line detail
3. Set `shift` to 3 (default) unless experimenting
4. Choose `random_seed: true` for exploration or `random_seed: false` with a specific seed for reproducibility
5. Record all parameters for documentation

**Got:** A complete parameter set ready for generation: resolution, steps, shift, seed strategy.

**If fail:** If unsure about resolution, default to 1024x1024 (1:1) — it works for most ornamental contexts and is the fastest to generate.

### Step 5: Generate Image

Invoke the Z-Image MCP tool to produce the ornament.

1. Call `mcp__hf-mcp-server__gr1_z_image_turbo_generate` with:
   - `prompt`: the constructed prompt from Step 3
   - `resolution`: from Step 4
   - `steps`: from Step 4
   - `shift`: from Step 4
   - `random_seed`: from Step 4
   - `seed`: specific seed if `random_seed` is false
2. Record the returned seed value for reproducibility
3. Note the generation time

**Got:** A generated image and a seed value. The image shows recognizable ornamental forms in monochrome.

**If fail:** If the MCP tool is unavailable, verify that hf-mcp-server is configured (see `configure-mcp-server` or `troubleshoot-mcp-connection`). If the tool is available but returns an error, simplify the prompt and retry. If the generated image is entirely abstract with no ornamental character, the prompt needs more specific structural language — return to Step 3.

### Step 6: Evaluate Against Style Criteria

Assess the generated image against four criteria.

```
Monochrome Ornament Evaluation Rubric:
┌─────────────────────┬───────────────────────────────────────────────────────┐
│ Criterion           │ Evaluation Questions                                  │
├─────────────────────┼───────────────────────────────────────────────────────┤
│ 1. Symmetry         │ Does the design exhibit the intended symmetry type?   │
│                     │ Is it visually balanced? Are repeating elements       │
│                     │ consistent?                                           │
├─────────────────────┼───────────────────────────────────────────────────────┤
│ 2. Monochrome       │ Is the image truly black and white? Are there         │
│    Fidelity         │ unwanted grays, colors, or gradients? Does the        │
│                     │ rendering style match the request?                    │
├─────────────────────┼───────────────────────────────────────────────────────┤
│ 3. Period Accuracy  │ Would this design be recognizable as belonging to     │
│                     │ the specified period? Are the motifs period-           │
│                     │ appropriate? Does it avoid anachronistic elements?    │
├─────────────────────┼───────────────────────────────────────────────────────┤
│ 4. Detail Level     │ Is the level of detail appropriate for the rendering  │
│                     │ style? Line art should have clean lines; woodcut      │
│                     │ should show bold strokes; engraving should show       │
│                     │ systematic hatching.                                  │
└─────────────────────┴───────────────────────────────────────────────────────┘
```

1. Score each criterion: **Strong** (clearly meets), **Adequate** (partially meets), **Weak** (does not meet)
2. Note specific observations for each criterion
3. If 3+ criteria score Strong, the design is successful
4. If 2+ criteria score Weak, return to Step 3 for prompt refinement

**Got:** A scored evaluation with specific observations. Most first-generation images will score Adequate on 2-3 criteria.

**If fail:** If all criteria score Weak, the prompt may be too abstract or too complex. Simplify to the most essential elements: one motif, one rendering style, explicit "monochrome black and white" constraint. Consider switching to a period with higher mono suitability.

### Step 7: Iterate or Finalize

Refine the design through targeted iteration or accept the result.

**Iteration Strategies:**
1. **Seed-locked refinement**: Keep the same seed, adjust the prompt slightly — this evolves the composition while maintaining its basic structure
2. **Random exploration**: Use `random_seed: true` with the same prompt — this produces variations on the same concept
3. **Prompt evolution**: Modify specific elements (change rendering style, add/remove motif details, adjust composition)

**Iteration Budget:** Limit to 3 iterations per design concept. If the result is not satisfactory after 3 iterations, reconsider the period/motif combination or rendering style fundamentally.

1. If the evaluation in Step 6 indicates specific weaknesses, adjust the prompt to address them:
   - Weak symmetry → add "perfectly symmetrical" or "mirror symmetry"
   - Color leaking → add "pure black and white, no gray tones, no color"
   - Wrong period feel → add specific period reference artists or monuments
   - Insufficient detail → increase steps to 10-12, add "highly detailed"
2. Regenerate using Step 5
3. Re-evaluate using Step 6
4. Accept when 3+ criteria score Strong or iteration budget is exhausted

**Got:** An improved image after 1-2 iterations, or a decision to accept the current best result.

**If fail:** If iteration is not improving results, the fundamental prompt concept may not translate well to the model. Try a different motif from the same period, or switch the rendering style entirely (e.g., from line art to silhouette).

### Step 8: Document the Design

Create a complete record of the final design for reproducibility and reference.

1. Record the following:
   - **Period**: Historical period name and date range
   - **Motif**: Primary motif(s) used
   - **Rendering Style**: Line art, silhouette, woodcut, etc.
   - **Final Prompt**: The exact prompt that produced the accepted image
   - **Seed**: The seed value for reproduction
   - **Resolution**: The resolution used
   - **Steps/Shift**: Generation parameters
   - **Evaluation**: Brief notes on the four criteria scores
   - **Iterations**: Number of iterations and key changes made
2. Note any art historical observations — how the generated design compares to historical examples
3. Suggest potential applications: print, digital border, textile pattern, etc.

**Got:** A reproducible record that allows the exact image to be regenerated and its design lineage understood.

**If fail:** If documentation feels excessive, at minimum record the final prompt and seed — these two values are sufficient to reproduce the image.

## Key Motifs Reference

The following motifs appear across multiple historical periods and form the core vocabulary of classical ornament:

- **Acanthus**: Deeply lobed leaf; Greek origin, dominant in Roman and Renaissance ornament
- **Palmette**: Fan-shaped leaf cluster; Egyptian and Greek, ancestor of the anthemion
- **Anthemion**: Alternating palmette-and-lotus frieze; Greek, endlessly adapted
- **Guilloche**: Interlocking circles forming a chain; ancient, universal
- **Meander / Greek Key**: Angular spiral forming a continuous band; quintessentially Greek
- **Arabesque**: Infinitely extending vegetal scroll; Islamic, non-representational by principle
- **Trefoil / Quatrefoil**: Three/four-lobed forms within a circle; Gothic tracery
- **Rosette**: Radially symmetric flower form; universal across all periods
- **Scroll (C and S)**: Spiraling forms; Baroque and Rococo signature elements
- **Grotesque**: Fantastical human-animal-vegetal hybrid; Roman, revived in Renaissance
- **Interlace / Knotwork**: Woven bands without beginning or end; Celtic, Islamic, Byzantine
- **Lotus**: Stylized water lily; Egyptian origin, spread across Asian ornament traditions

## Validation

- [ ] A specific historical period was selected with rationale
- [ ] Motif structure was analyzed (symmetry, scaffold, fill, edge treatment)
- [ ] Prompt includes explicit monochrome constraint ("black and white" or equivalent)
- [ ] Prompt specifies a rendering style (line art, silhouette, woodcut, etc.)
- [ ] Resolution matches the application context
- [ ] Generated image was evaluated against the 4-point rubric
- [ ] Seed value was recorded for reproducibility
- [ ] Final design is documented with prompt, seed, and parameters

## Pitfalls

- **Omitting the monochrome constraint**: Z-Image defaults to color. Without explicit "monochrome, black and white" in the prompt, you will get color output. Add the constraint early in the prompt, not as an afterthought
- **Over-specifying the prompt**: Prompts over 50 words tend to produce confused results. Keep to one motif, one rendering style, one composition type. Quality comes from clarity, not quantity
- **Ignoring period grammar**: Each period has structural rules. Gothic trefoils inside Egyptian frames, or Baroque scrolls in Greek meander borders, produce visual incoherence. Stay within the period vocabulary
- **Expecting vector output**: Z-Image produces raster images. For true vector line art, the generated image serves as a reference for manual tracing, not a final production asset
- **Skipping the structural analysis**: Jumping from period selection to prompt without analyzing motif structure produces generic "decorative" results rather than historically grounded ornament

## Related Skills

- `ornament-style-color` — the polychromatic companion to this skill; adds color palette definition and color-to-structure mapping
- `meditate` — focused attention and visual imagination practices can inform ornamental composition
- `review-web-design` — design review principles (visual hierarchy, rhythm, balance) apply directly to ornamental composition

Related Skills

ornament-style-modern

9
from pjt222/agent-almanac

Design ornamental patterns using modern and speculative aesthetics with colorblind-accessible color scales. Breaks free from historical period constraints to explore cyberpunk, solarpunk, biopunk, brutalist, vaporwave, and other contemporary genres. Includes CVD (Color Vision Deficiency) awareness and perceptually uniform scales (viridis, cividis, inferno). Use when creating ornamental designs in modern or genre-specific aesthetics, designing patterns that must be colorblind-accessible, or exploring hybrid motifs combining historical ornament with contemporary visual language.

ornament-style-color

9
from pjt222/agent-almanac

Design polychromatic ornamental patterns grounded in Alexander Speltz's classical ornament taxonomy. Builds on monochrome structural analysis by adding period-authentic color palettes, color-to-motif mapping, and rendering styles suited to painted, illuminated, and glazed ornament. Use when creating decorative designs where color is integral to the tradition (Islamic tilework, illuminated manuscripts, Art Nouveau), exploring how historical periods used color in ornament, or producing colored reference imagery for design, illustration, or educational materials.

skill-name-here

9
from pjt222/agent-almanac

One to three sentences describing what this skill accomplishes, followed by key activation triggers. This field is the primary mechanism agents use to decide whether to activate the skill — it is read during discovery before the full body is loaded. Start with a verb. Include the most important "when to use" conditions inline. Max 1024 characters.

write-vignette

9
from pjt222/agent-almanac

Create R package vignettes using R Markdown or Quarto. Covers vignette setup, YAML configuration, code chunk options, building and testing, and CRAN requirements for vignettes. Use when adding a Getting Started tutorial, documenting complex workflows spanning multiple functions, creating domain-specific guides, or when CRAN submission requires user-facing documentation beyond function help pages.

write-validation-documentation

9
from pjt222/agent-almanac

Write IQ/OQ/PQ validation documentation for computerized systems in regulated environments. Covers protocols, reports, test scripts, deviation handling, and approval workflows. Use when validating R or other software for regulated use, preparing for a regulatory audit, documenting qualification of computing environments, or creating and updating validation protocols and reports for new or re-qualified systems.

write-testthat-tests

9
from pjt222/agent-almanac

Write comprehensive testthat (edition 3) tests for R package functions. Covers test organization, expectations, fixtures, mocking, snapshot tests, parameterized tests, and achieving high coverage. Use when adding tests for new package functions, increasing test coverage for existing code, writing regression tests for bug fixes, or setting up test infrastructure for a package that lacks it.

write-standard-operating-procedure

9
from pjt222/agent-almanac

Write a GxP-compliant Standard Operating Procedure (SOP). Covers regulatory SOP template structure (purpose, scope, definitions, responsibilities, procedure, references, revision history), approval workflow design, periodic review scheduling, and operational procedures for system use. Use when a new validated system requires operational procedures, when existing informal procedures need formalisation, when an audit finding cites missing procedures, when a change control triggers SOP updates, or when periodic review identifies outdated procedural content.

write-roxygen-docs

9
from pjt222/agent-almanac

Write roxygen2 documentation for R package functions, datasets, and classes. Covers all standard tags, cross-references, examples, and generating NAMESPACE entries. Follows tidyverse documentation style. Use when adding documentation to new exported functions, documenting internal helpers or datasets, documenting S3/S4/R6 classes and methods, or fixing documentation-related R CMD check notes.

write-incident-runbook

9
from pjt222/agent-almanac

Create structured incident runbooks with diagnostic steps, resolution procedures, escalation paths, and communication templates for effective incident response. Use when documenting response procedures for recurring alerts, standardizing incident response across an on-call rotation, reducing MTTR with clear diagnostic steps, creating training materials for new team members, or linking alert annotations directly to resolution procedures.

write-helm-chart

9
from pjt222/agent-almanac

Create production-ready Helm charts for Kubernetes application deployment with templating, values management, chart dependencies, hooks, and testing. Covers chart structure, Go template syntax, values.yaml design, chart repositories, versioning, and best practices for maintainable and reusable charts. Use when packaging a Kubernetes application for repeatable deployments, parameterizing manifests for multiple environments, managing complex multi-component applications with dependencies, or standardizing deployment practices with versioned rollback capability across teams.

write-continue-here

9
from pjt222/agent-almanac

Write a CONTINUE_HERE.md file capturing current session state so a fresh Claude Code session can pick up where this one left off. Covers assessing recent work, structuring the continuation file with objective, completed, in-progress, next-steps, and context sections, and verifying the file is actionable. Use when ending a session with unfinished work, handing off context between sessions, or preserving task state that git alone cannot capture.

write-claude-md

9
from pjt222/agent-almanac

Create an effective CLAUDE.md file that provides project-specific instructions to AI coding assistants. Covers structure, common sections, do/don't patterns, and integration with MCP servers and agent definitions. Use when starting a new project where AI assistants will be used, improving AI behavior on an existing project, documenting project conventions and constraints, or integrating MCP servers or agent definitions into a project workflow.