Global Settings Reference¶
This document provides a comprehensive reference for rumdl's global configuration settings. Global settings control rumdl's overall behavior and apply to all rules and operations.
Overview¶
Global settings are configured in the [global] section of your configuration file (.rumdl.toml or
pyproject.toml). These settings control file selection, rule enablement, and general linting behavior.
Tip: In
.rumdl.toml, global keys can also be placed at the top level without a[global]section header, similar to ruff.toml. If both top-level and[global]keys exist, the[global]section takes precedence. See Top-level shorthand for details.
Quick Reference¶
| Setting | Type | Default | Description |
|---|---|---|---|
extends |
string |
not set | Inherit settings from another config file |
enable |
string[] |
not set | Enable only specific rules |
disable |
string[] |
[] |
Disable specific rules |
extend-enable |
string[] |
[] |
Additional rules to enable (additive) |
extend-disable |
string[] |
[] |
Additional rules to disable (additive) |
per-file-ignores |
table |
{} |
Disable specific rules for specific files |
exclude |
string[] |
[] |
Files/directories to exclude |
include |
string[] |
[] |
Files/directories to include |
respect_gitignore |
boolean |
true |
Respect .gitignore files |
line_length |
integer |
80 |
Default line length for rules |
flavor |
string |
"standard" |
Markdown flavor to use |
per-file-flavor |
table |
{} |
Per-file flavor overrides |
output-format |
string |
"text" |
Output format for linting results |
cache |
boolean |
true |
Enable result caching |
cache_dir |
string |
.rumdl_cache |
Directory for cache files |
Configuration Examples¶
TOML Configuration (.rumdl.toml)¶
[global]
# Disable specific rules
disable = ["MD013", "MD033"]
# Add opt-in rules on top of defaults (additive)
extend-enable = ["MD060", "MD063"]
# Exclude files and directories
exclude = [
"node_modules",
"build",
"dist",
"*.tmp.md",
"docs/generated/**"
]
# Include only specific files
include = [
"README.md",
"docs/**/*.md",
"**/*.markdown"
]
# Don't respect .gitignore files
respect_gitignore = false
# Set global line length (used by MD013 and other line-length rules)
line_length = 120
# Set markdown flavor (standard, mkdocs)
flavor = "mkdocs"
# Per-file flavor overrides (pattern → flavor)
[per-file-flavor]
"**/*.mdx" = "mdx"
"notebooks/**/*.qmd" = "quarto"
# Disable specific rules for specific files
[per-file-ignores]
"README.md" = ["MD033"] # Allow inline HTML in README
"docs/api/**/*.md" = ["MD013", "MD041"] # Generated API docs
"SUMMARY.md" = ["MD025"] # MkDocs/mdBook table of contents
pyproject.toml Configuration¶
[tool.rumdl]
# Global options at root level (both snake_case and kebab-case supported)
disable = ["MD013", "MD033"]
extend-enable = ["MD060"]
exclude = ["node_modules", "build", "dist"]
include = ["docs/*.md", "README.md"]
respect_gitignore = true
line_length = 120
flavor = "standard"
# Per-file rule ignores (both snake_case and kebab-case supported)
[tool.rumdl.per-file-ignores]
"README.md" = ["MD033"]
"SUMMARY.md" = ["MD025"]
Top-level shorthand¶
In .rumdl.toml, you can place global keys at the top level without wrapping them in a [global] section.
This is convenient for simple configs and matches the convention used by ruff.toml:
line-length = 88
disable = []
exclude = []
respect-gitignore = true
[MD004]
style = "dash"
[MD013]
line_length = 88
code_blocks = false
This is equivalent to:
[global]
line-length = 88
disable = []
exclude = []
respect-gitignore = true
[MD004]
style = "dash"
[MD013]
line_length = 88
code_blocks = false
Precedence: If the same key appears both at the top level and inside [global], the [global] value wins.
Supported keys: enable, disable, include, exclude, extend-enable, extend-disable,
respect-gitignore, force-exclude, line-length, output-format, cache-dir, cache, fixable,
unfixable, flavor.
Notes:
- Table-type settings (
per-file-ignores,per-file-flavor,code-block-tools) still require their own section headers. Only scalar and array keys can be placed at the top level. - This shorthand is only available in
.rumdl.toml/rumdl.toml. Inpyproject.toml, use[tool.rumdl]as shown in the pyproject.toml example above.
Rule Selection Model¶
rumdl uses four settings to control which rules are active. These follow the same model as Ruff's lint rule selection:
| rumdl | Ruff equivalent | CLI flag | Behavior |
|---|---|---|---|
enable |
select |
--enable |
Set the enabled rules |
disable |
ignore |
--disable |
Set the disabled rules |
extend-enable |
extend-select |
— | Add rules to the enabled set |
extend-disable |
extend-ignore |
— | Add rules to the disabled set |
How rules are resolved¶
- Start with the base set. If
enableis omitted, the base set is all rules except opt-in rules. Ifenableis set, only the listed rules form the base set. - Merge
extend-enable. Any rules listed inextend-enableare added to the base set. This is the way to activate opt-in rules without replacing the entire default set. - Apply
disableandextend-disable. Rules in either list are removed. Disabling always wins over enabling.
Key behaviors:
enableomitted — all non-opt-in rules are active (the default)enable = [](empty list) — no rules are active; this is an explicit empty allowlist, not the same as omittingenableenable = ["ALL"]— every rule is active, including opt-in rulesextend-enable = ["ALL"]— every rule is active (same effect asenable = ["ALL"])disable = ["all"]with noenable— no rules are active- Disabling wins — if a rule appears in both an enable list and a disable list, it is disabled
Config precedence and merging¶
rumdl loads configuration from config files (with per-directory resolution when available), then applies CLI overrides on top. When CLI flags overlap with file settings:
--enableand--disablereplace the config file value entirely.extend-enableandextend-disable(config file only) merge additively with the baseenable/disablevalues.
# Config file has disable = ["MD013"]
# CLI replaces it — only MD033 is disabled, MD013 is re-enabled
rumdl check --disable MD033 .
To disable MD033 in addition to whatever the config file disables, use extend-disable in the config file instead of CLI --disable.
Opt-in rules¶
Some rules are excluded from the default set because they are opinionated, project-specific,
or may require configuration.
These rules must be explicitly activated via extend-enable or enable = ["ALL"].
Current opt-in rules:
| Rule | Description |
|---|---|
| MD060 | Table column formatting |
| MD063 | Heading capitalization |
| MD070 | Nested code fence collision |
| MD072 | Frontmatter key sort order |
| MD073 | Table of contents validation |
| MD074 | MkDocs nav validation |
Common patterns¶
Use only a handful of rules (strict allowlist):
Use all defaults but disable a few:
Use all defaults plus opt-in rules:
Use every rule including opt-in, minus a few:
Detailed Settings Reference¶
extends¶
Type: string
Default: not set
CLI Equivalent: None (configuration file only)
Specifies a base configuration file to inherit settings from. The current config file's settings are merged on top of the base config.
This is a top-level key (not inside [global]).
In .rumdl.toml:
In pyproject.toml:
Path Resolution:
- Relative paths are resolved relative to the config file's directory (not the working directory)
~/prefix expands to the user's home directory- Absolute paths are used as-is
- The extended file can be
.rumdl.toml,rumdl.toml, orpyproject.toml
Merge Behavior:
When config B extends config A:
- Replace fields (
enable,disable,line-length,flavor,exclude, etc.): B's value replaces A's if B specifies it - Union fields (
extend-enable,extend-disable): B's values accumulate with A's - Unspecified fields: A's values are kept
- Rule-specific settings: B's
[MD007]overrides A's[MD007](per-key)
Chains:
Configs can chain: A extends B extends C. The base config is loaded first recursively, then each child merges on top. Maximum chain depth is 10.
Circular Detection:
Circular references (A extends B extends A) are detected and produce a clear error.
Common Patterns:
Subdirectory override:
# docs/.rumdl.toml — relaxed rules for documentation
extends = "../.rumdl.toml"
[global]
extend-disable = ["MD013"]
Shared base config:
# .rumdl.toml in each project
extends = "~/.config/rumdl/base.rumdl.toml"
[global]
disable = ["MD033"]
enable¶
Type: string[]
Default: not set (all rules enabled; disable applies normally)
CLI Equivalent: --enable
Enables only the specified rules. When this option is set, all other rules are disabled except those explicitly listed.
Usage Notes:
- Rule IDs are case-insensitive but conventionally uppercase (e.g., "MD001")
enable = [](empty list) disables all rules — nothing will be linted- Omitting
enableentirely uses the default: all non-opt-in rules enabled,disableapplied normally enable = ["ALL"]explicitly enables every rule including opt-in rules;disablestill applies on top- If
enablelists specific rules, only those rules run (subject todisableandextend-disable) - When
enableis set,extend-enableentries are merged into the allowlist - To add opt-in rules without replacing the defaults, use
extend-enableinstead
Example CLI usage:
extend-enable¶
Type: string[]
Default: [] (no additional rules)
CLI Equivalent: None (configuration file only)
Adds rules to the enabled set. This is the primary way to activate opt-in rules (MD060, MD063, MD072, MD073, MD074) without replacing the default rule set.
When CLI --enable is used, it replaces the config file's enable list entirely.
extend-enable in the config file is always additive —
see Config precedence and merging.
Usage Notes:
extend-enable = ["ALL"]enables every rule including opt-in rules (same asenable = ["ALL"])- Can be combined with
enable— when both are set,extend-enableentries are merged into theenableallowlist - Disabling always wins: if a rule appears in both
extend-enableanddisable/extend-disable, it is disabled - Rule IDs are case-insensitive but conventionally uppercase
Example: Enable opt-in rules alongside defaults:
Example: Combine with disable:
[global]
extend-enable = ["MD060"] # Add table formatting
disable = ["MD013"] # Remove line length checking
disable¶
Type: string[]
Default: [] (no rules disabled)
CLI Equivalent: --disable
Disables the specified rules. All other rules remain enabled.
Usage Notes:
- Rule IDs are case-insensitive but conventionally uppercase
disable = ["all"]disables every rule; only rules listed inenable(if set) are active- Commonly disabled rules include:
MD013: Line length (for projects with longer lines)MD033: Inline HTML (for projects that use HTML in Markdown)MD041: First line heading (for files that don't start with headings)- Disabling always wins over enabling — if a rule appears in both
enableanddisable, it is disabled - CLI
--disablereplaces the config file'sdisablelist; useextend-disablein the config file for values that shouldn't be overridden by CLI
Example CLI usage:
extend-disable¶
Type: string[]
Default: [] (no additional rules disabled)
CLI Equivalent: None (configuration file only)
Adds rules to the disabled set. Always merges additively with disable, regardless of source.
Useful for adding disabled rules that survive CLI --disable overrides —
see Config precedence and merging.
Usage Notes:
extend-disable = ["ALL"]disables every rule (effectively stops all linting)- Disabling always wins over enabling: if a rule appears in both
extend-enableandextend-disable, it is disabled - Rule IDs are case-insensitive but conventionally uppercase
Example: Disable rules that survive CLI overrides:
[global]
disable = ["MD013"]
extend-disable = ["MD033"] # Always disabled, even if --disable overrides the list above
With rumdl check --disable MD041 ., the result is: MD013 replaced by MD041 (from CLI), plus MD033 (from extend-disable, always additive). Final disabled set: MD041, MD033.
per-file-ignores¶
Type: table (file patterns mapped to rule arrays)
Default: {} (no per-file ignores)
CLI Equivalent: None (configuration file only)
Disables specific rules for specific files or file patterns. This is useful when certain files have different requirements than the rest of your documentation.
[per-file-ignores]
# Disable inline HTML check for GitHub README (often has badges)
"README.md" = ["MD033"]
# Disable line length and first-line heading for generated API docs
"docs/api/**/*.md" = ["MD013", "MD041"]
# Allow multiple top-level headings in table of contents files
"SUMMARY.md" = ["MD025"]
"**/TOC.md" = ["MD025"]
# Disable heading style check for legacy documentation
"docs/legacy/**/*.md" = ["MD003", "MD022"]
Pattern Syntax:
Patterns use standard glob syntax:
*matches any characters except path separators**matches any characters including path separators (recursive)?matches a single character{a,b}matches eitheraorb(brace expansion)
To match multiple specific files, use brace expansion:
[per-file-ignores]
# Match both AGENTS.md and README.md
"{AGENTS.md,README.md}" = ["MD033"]
# Match multiple directories
"{docs,guides}/**/*.md" = ["MD013"]
Alternatively, use separate entries (more verbose but equivalent):
Note: Commas are literal characters in glob patterns. The pattern
"A.md,B.md"matches a file literally namedA.md,B.md, not two separate files. Use"{A.md,B.md}"to match multiple files.
Usage Notes:
- Rule IDs are case-insensitive but conventionally uppercase
- More specific patterns take precedence over general ones
- Useful for handling special files like:
README.mdfiles (often contain badges, HTML, custom formatting)- Generated documentation (API docs, changelogs)
- Table of contents files (
SUMMARY.md,TOC.md) - Legacy or third-party documentation
- Combines with global
enable/disablesettings
Behavior:
- Global rules are applied first
- Per-file ignores override global settings for matching files
- If a file matches multiple patterns, all ignores are combined
Example with precedence:
[global]
disable = ["MD013"] # Disable line length globally
[per-file-ignores]
"README.md" = ["MD033", "MD041"] # Also disable HTML and first-line heading for README
"docs/**/*.md" = ["MD033"] # Allow HTML in docs
Result:
README.md: MD013, MD033, MD041 disableddocs/guide.md: MD013, MD033 disabledother.md: MD013 disabled
Common Use Cases:
-
MkDocs/mdBook Projects:
-
GitHub Projects with Badges:
-
Generated Documentation:
-
Mixed Documentation Sources:
-
Documentation Generators with HTML Links:
For documentation generators (mdBook, Jekyll, Hugo) that compile markdown to HTML and place sources in different locations:
[per-file-ignores] # mdBook projects - HTML links in book/ point to book/src/*.md sources "book/**/*.md" = ["MD057"] # Jekyll projects - HTML links in _posts/ point to generated files "_posts/**/*.md" = ["MD057"] "_docs/**/*.md" = ["MD057"] # Hugo projects - HTML links in content/ point to generated files "content/**/*.md" = ["MD057"]See MD057 documentation for more details.
exclude¶
Type: string[]
Default: [] (no files excluded)
CLI Equivalent: --exclude
Specifies files and directories to exclude from linting. Supports glob patterns.
[global]
exclude = [
"node_modules", # Exclude entire directory
"build/**", # Exclude directory and all subdirectories
"*.tmp.md", # Exclude files with specific pattern
"docs/generated/**", # Exclude generated documentation
".git", # Exclude version control directory
"vendor/", # Exclude third-party code
]
Supported Patterns:
directory/- Exclude entire directory**/*.ext- Exclude all files with extension in any subdirectory*.pattern- Exclude files matching pattern in current directorypath/**/file- Exclude specific files in any subdirectory of path
Usage Notes:
- Patterns are relative to the project root
- Exclude patterns are processed before include patterns
- More specific patterns take precedence over general ones
- Useful for excluding generated files, dependencies, and temporary files
Example CLI usage:
include¶
Type: string[]
Default: [] (all Markdown files included)
CLI Equivalent: --include
Specifies files and directories to include in linting. When set, only matching files are processed.
[global]
include = [
"README.md", # Include specific file
"docs/**/*.md", # Include all .md files in docs/
"**/*.markdown", # Include all .markdown files
"CHANGELOG.md", # Include specific files
"src/**/*.md", # Include documentation in source
]
Usage Notes:
- If
includeis empty, all Markdown files are included (subject to exclude patterns) - When
includeis specified, only matching files are processed - Combine with
excludefor fine-grained control - Useful for limiting linting to specific documentation areas
Example CLI usage:
respect_gitignore¶
Type: boolean
Default: true
CLI Equivalent: --respect-gitignore / --respect-gitignore=false
Controls whether rumdl respects .gitignore files when scanning for Markdown files.
[global]
respect_gitignore = true # Default: respect .gitignore
# or
respect_gitignore = false # Ignore .gitignore files
Behavior:
true(default): Files and directories listed in ignore files are automatically excludedfalse: Ignore files are not considered, all Markdown files are scanned
Supported ignore files:
.gitignore- Standard Git ignore patterns.ignore- Additional ignore patterns (used by ripgrep, fd, and other tools)
Both file types use the same gitignore pattern syntax and are respected at any level in the directory tree.
Usage Notes:
- This setting only affects directory scanning, not explicitly provided file paths
- Useful for linting files that are normally ignored (e.g., generated docs)
- When disabled, you may need more specific
excludepatterns - Use
.ignorefor rumdl-specific exclusions without affecting Git
Example CLI usage:
line_length¶
Type: integer
Default: 80
CLI Equivalent: None (rule-specific only)
Sets the global default line length used by rules that check line length (primarily MD013). Rules can override this value with their own configuration.
Behavior:
- Used as the default line length for MD013 and other line-length-related rules
- Rule-specific configurations override the global setting
- Useful for projects that want a consistent line length across all line-length rules
Usage Notes:
- Must be a positive integer
- Common values: 80 (traditional), 100 (relaxed), 120 (modern)
- Individual rules can still override this setting in their own configuration
- When importing from markdownlint configs, top-level
line-lengthis mapped to this setting
Example with rule override:
[global]
line_length = 100 # Global default
[MD013]
line_length = 120 # MD013 uses 120, overriding global setting
flavor¶
Type: string
Default: "standard"
CLI Equivalent: --flavor
Specifies the Markdown flavor to use for parsing and linting. Different flavors have different parsing rules and feature support.
Available Flavors:
"standard"(default): CommonMark 0.31.2 + GFM extensions (tables, task lists, strikethrough, autolinks)"gfm": GitHub Flavored Markdown with security-sensitive HTML warnings and extended autolinks"mkdocs": MkDocs-specific extensions (admonitions, content tabs, autorefs, mkdocstrings)"mdx": MDX with JSX components, attributes, expressions, and ESM imports"quarto": Quarto/RMarkdown for scientific publishing (citations, shortcodes, div blocks)
Aliases: "commonmark" is an alias for "standard", "github" is an alias for "gfm"
Behavior:
- The
standardflavor is based on CommonMark 0.31.2 with widely-adopted GFM extensions enabled by default - Each flavor adjusts specific rule behavior where that system differs from standard Markdown
- See Flavors Overview for detailed rule adjustments per flavor
Usage Notes:
- Choose the flavor that matches your documentation system
- Use
standardfor generic Markdown or when you want the strictest linting - Use
gfmfor GitHub-hosted documentation with security-conscious HTML handling - Use
mkdocsfor MkDocs or Material for MkDocs projects - Use
mdxfor React/Next.js documentation with JSX components - Use
quartofor scientific documents with R/Python code execution
Example CLI usage:
per-file-flavor¶
Type: table (file patterns mapped to flavors)
Default: {} (no per-file overrides)
CLI Equivalent: None (configuration file only)
Specifies Markdown flavors for specific files or file patterns. This allows different parts of your project to use different Markdown dialects.
[per-file-flavor]
"docs/**/*.md" = "mkdocs"
"**/*.mdx" = "mdx"
"**/*.qmd" = "quarto"
"examples/**/*.md" = "standard"
Available Flavors:
"standard"(default): Standard Markdown with GFM extensions (tables, task lists, strikethrough)"gfm"or"github": Alias for standard (pulldown-cmark already supports GFM)"commonmark": Alias for standard"mkdocs": MkDocs-specific extensions (auto-references, admonitions)"mdx": MDX flavor with JSX and ESM support"quarto": Quarto/RMarkdown for scientific publishing
Behavior:
- Uses "first match wins" semantics - order matters in the configuration
- Patterns are matched against relative paths from the project root
- Falls back to global
flavorsetting if no pattern matches - Falls back to auto-detection by file extension if no global flavor is set
Pattern Syntax:
*matches any characters except path separators**matches any characters including path separators?matches a single character- Patterns are relative to the project root
Usage Notes:
- Useful for projects with mixed documentation (e.g., MkDocs site + MDX components)
- Order patterns from most specific to least specific
- Auto-detection works for common extensions:
.mdx→ MDX,.qmd/.Rmd→ Quarto
Example: Mixed Documentation Project:
[global]
flavor = "standard" # Default for files not matching any pattern
[per-file-flavor]
# MkDocs documentation
"docs/**/*.md" = "mkdocs"
# React components with MDX
"src/components/**/*.mdx" = "mdx"
# Jupyter/Quarto notebooks
"notebooks/**/*.qmd" = "quarto"
# Keep README and CHANGELOG as standard
"README.md" = "standard"
"CHANGELOG.md" = "standard"
Example: Monorepo with Multiple Doc Systems:
[per-file-flavor]
"packages/website/docs/**/*.md" = "mkdocs"
"packages/storybook/**/*.mdx" = "mdx"
"packages/api/docs/**/*.md" = "standard"
output-format¶
Type: string
Default: "text"
CLI Equivalent: --output-format
Environment Variable: RUMDL_OUTPUT_FORMAT
Specifies the output format for linting results.
Available Formats:
"text"(default): One line per warning with file, line, column, rule, and message"full": Source lines with caret underlines highlighting the exact violation location"concise": Minimal output (one line per warning, no brackets)"grouped": Warnings grouped by file with a header per file"json": JSON array of all warnings (collected across files)"json-lines": One JSON object per warning (streaming)"github": GitHub Actions annotation format (::warning/::error)"gitlab": GitLab Code Quality report (JSON)"pylint": Pylint-compatible format"azure": Azure Pipelines logging commands"sarif": SARIF 2.1.0 for static analysis tools"junit": JUnit XML for CI test reporters
Precedence:
- CLI flag (
--output-format) wins - Environment variable (
RUMDL_OUTPUT_FORMAT) overrides config - Config file setting (
output-format) - Default (
"text")
Usage Notes:
- Use
githubformat in GitHub Actions for inline annotations - Use
jsonorsariffor integration with other tools - The environment variable is useful for CI/CD pipelines where you want to override the project config
Example CLI usage:
# Use GitHub format for Actions
rumdl check --output-format github .
# Or via environment variable
RUMDL_OUTPUT_FORMAT=github rumdl check .
Example GitHub Actions workflow:
cache¶
Type: boolean
Default: true
CLI Equivalent: --no-cache (to disable)
Controls whether rumdl caches linting results to speed up subsequent runs.
Behavior:
true(default): Results are cached based on file content hashesfalse: Every run processes all files from scratch
Usage Notes:
- Caching significantly speeds up repeated linting of unchanged files
- Cache is automatically invalidated when file content changes
- Disable caching during development when debugging rule changes
- Use
--no-cacheCLI flag for one-time cache bypass without changing config
Example CLI usage:
cache_dir¶
Type: string
Default: .rumdl_cache
CLI Equivalent: --cache-dir
Environment Variable: RUMDL_CACHE_DIR
Specifies the directory where rumdl stores cache files.
[global]
cache_dir = ".rumdl_cache" # Default location
# or
cache_dir = "/tmp/rumdl-cache" # Custom location
Behavior:
- Cache files are stored in this directory
- Directory is created automatically if it doesn't exist
- Each file's cache entry is based on its content hash
Usage Notes:
- Default
.rumdl_cacheis relative to the current working directory - Use absolute paths for consistent behavior across different working directories
- Consider adding the cache directory to
.gitignore - Environment variable
RUMDL_CACHE_DIRcan override config file settings
Example CLI usage:
# Use custom cache directory
rumdl check --cache-dir /tmp/rumdl-cache .
# Or via environment variable
RUMDL_CACHE_DIR=/tmp/rumdl-cache rumdl check .
Adding to .gitignore:
Per-Directory Configuration¶
When running rumdl check . from the project root, rumdl discovers and applies
configuration files on a per-directory basis. Files in a subdirectory with their
own .rumdl.toml will use that config instead of the root config.
This follows the same model as Ruff's per-directory configuration.
How it works¶
- rumdl identifies the project root (the directory containing
.git) - For each file being linted, rumdl searches from the file's directory upward to the project root for the nearest config file
- Files are grouped by their effective config, and each group is linted with its own rules and settings
Config file search order¶
At each directory level, rumdl checks for config files in this order:
.rumdl.tomlrumdl.toml.config/rumdl.tomlpyproject.toml(only if it contains a[tool.rumdl]section)- Markdownlint config files (
.markdownlint.json,.markdownlint.yaml, etc.) as fallback
Subdirectory configs are standalone¶
Subdirectory configs are independent by default — they do not inherit from the root config. To inherit settings from a parent config, use extends:
project/
.rumdl.toml # line-length = 80
README.md # linted with line-length = 80
docs/
.rumdl.toml # standalone: only settings in this file apply
guide.md # linted with docs/.rumdl.toml
api/
endpoint.md # also linted with docs/.rumdl.toml (walks up to docs/)
To make a subdirectory config inherit from the root:
# docs/.rumdl.toml
extends = "../.rumdl.toml"
[global]
line-length = 120 # Override just this setting; inherit everything else
When per-directory resolution is active¶
Per-directory resolution only activates during auto-discovery mode. It is disabled when:
--config <file>is used (the explicit config applies to all files)--isolatedor--no-configis used (built-in defaults apply to all files)- No project root is found (no
.gitdirectory)
Example: monorepo with different standards¶
monorepo/
.rumdl.toml # strict defaults for most code
docs/
.rumdl.toml # relaxed rules for user-facing docs
getting-started.md
reference/
api.md # inherits docs/.rumdl.toml
blog/
.rumdl.toml # different style for blog posts
2024-01-post.md
src/
README.md # uses root .rumdl.toml (no subdirectory config)
# docs/.rumdl.toml - relaxed for documentation
extends = "../.rumdl.toml"
[global]
line-length = 120
extend-disable = ["MD013"]
# blog/.rumdl.toml - standalone config for blog
[global]
line-length = 100
disable = ["MD033", "MD041"]
Configuration Precedence¶
Settings are applied in the following order (later sources override earlier ones):
- Built-in defaults
- Configuration file (per-directory
.rumdl.tomlorpyproject.toml) - Command-line arguments
Example: Precedence in Action¶
Given this configuration file:
And this command:
The final configuration will be:
disable:["MD001", "MD013"](CLI overrides file)exclude:["temp/**"](CLI overrides file)- Paths:
["docs/"](CLI argument)
File Selection Logic¶
rumdl processes files using the following logic:
-
Start with candidate files:
- If paths are provided via CLI: use those files/directories
- Otherwise: recursively scan current directory for
.mdand.markdownfiles
-
Apply .gitignore filtering (if
respect_gitignore = true):- Skip files/directories listed in
.gitignorefiles
- Skip files/directories listed in
-
Apply include patterns (if specified):
- Keep only files matching at least one include pattern
-
Apply exclude patterns:
- Remove files matching any exclude pattern
-
Apply rule filtering:
- Process remaining files with enabled rules only
Example: File Selection¶
Given this configuration:
[global]
include = ["docs/**/*.md", "README.md"]
exclude = ["docs/temp/**", "*.draft.md"]
respect_gitignore = true
File selection process:
- Start with:
docs/guide.md,docs/temp/test.md,README.md,notes.draft.md - Apply includes:
docs/guide.md,docs/temp/test.md,README.md(notes.draft.md excluded) - Apply excludes:
docs/guide.md,README.md(docs/temp/test.md excluded) - Final files:
docs/guide.md,README.md
Common Configuration Patterns¶
Strict Documentation Projects¶
For projects with strict documentation standards:
[global]
# Enable only essential rules
enable = [
"MD001", # Heading increment
"MD003", # Heading style
"MD022", # Blanks around headings
"MD025", # Single title
"MD032", # Blanks around lists
]
# Focus on documentation directories
include = [
"README.md",
"CHANGELOG.md",
"docs/**/*.md",
"guides/**/*.md",
]
# Set consistent line length for documentation
line_length = 100
# Exclude generated or temporary content
exclude = [
"node_modules/**",
"build/**",
"docs/api/**", # Generated API docs
"*.draft.md", # Draft documents
]
Legacy Project Migration¶
For gradually adopting rumdl in existing projects:
[global]
# Start with basic rules only
enable = [
"MD001", # Heading increment
"MD022", # Blanks around headings
"MD025", # Single title
]
# Exclude problematic areas initially
exclude = [
"legacy-docs/**",
"third-party/**",
"vendor/**",
]
# Process only main documentation
include = [
"README.md",
"docs/user-guide/**/*.md",
]
Open Source Projects¶
For open source projects with community contributions:
[global]
# Disable rules that might be too strict for contributors
disable = [
"MD013", # Line length (can be restrictive)
"MD033", # Inline HTML (often needed for badges/formatting)
"MD041", # First line heading (README might start with badges)
]
# Include all documentation but exclude dependencies
include = ["**/*.md"]
exclude = [
"node_modules/**",
"vendor/**",
".git/**",
"coverage/**",
]
Development Workflow¶
For active development with frequent documentation changes:
[global]
# Enable rules that help maintain consistency
enable = [
"MD003", # Heading style
"MD004", # List style
"MD022", # Blanks around headings
"MD032", # Blanks around lists
"MD047", # Single trailing newline
]
# Include work-in-progress docs but exclude temp files
include = ["docs/**/*.md", "*.md"]
exclude = [
"*.tmp.md",
"*.draft.md",
".backup/**",
"node_modules/**",
]
# Don't respect .gitignore to catch uncommitted docs
respect_gitignore = false
Validation and Debugging¶
View Effective Configuration¶
To see how your global settings are applied:
# View full effective configuration
rumdl config
# View only global settings
rumdl config get global
# View specific global setting
rumdl config get global.exclude
Test File Selection¶
To see which files would be processed:
# Dry run with verbose output
rumdl check --verbose --dry-run .
# Or use a simple script to list files
find . -name "*.md" -o -name "*.markdown" | head -10
Common Configuration Issues¶
- No files found: Check your
include/excludepatterns andrespect_gitignoresetting - Too many files: Add more specific
excludepatterns or limitincludepatterns - Rules not applying: Verify rule names in
enable/disablelists (case-insensitive but check spelling) - Performance issues: Exclude large directories like
node_modules,vendor, or build outputs
Integration with CI/CD¶
GitHub Actions¶
- name: Lint Markdown
run: |
rumdl check --output json . > lint-results.json
rumdl check . # Also show human-readable output
Pre-commit Hook¶
- repo: https://github.com/rvben/rumdl-pre-commit
rev: v0.0.200
hooks:
- id: rumdl
args: [--config=.rumdl.toml]
See Also¶
- Configuration Guide - Basic configuration setup
- Rules Reference - Complete list of available rules
- CLI Reference - Command-line options
- Rule-specific Configuration - Configuring individual rules
- Code Block Tools - External linters/formatters for code blocks [preview]