ryl

ryl is a fast YAML linter (Rust, yamllint-compatible) built for programmatic use: stable exit codes, machine-readable output, in-place fixes, and stdin support.

Install: pip install ryl, cargo install ryl, or npm i -g @owenlamont/ryl. Pre-commit via the owenlamont/ryl-pre-commit hook (ryl for YAML files, ryl-markdown for embedded YAML).

Critical: ryl has no default-on rules

ryl never enables a rule unless a configuration turns it on. When it has a file to lint but no config (or a config that enables zero rules) it exits 2, not 0 (an empty or fully-ignored input set still exits 0). This is stricter than yamllint. Always give it a config first:

ryl check -d 'extends: default' .          # quick: yamllint's standard rule set

Or commit a ryl.toml / .ryl.toml (in pyproject.toml, prefix the tables with tool.ryl, so [rules] becomes [tool.ryl.rules]):

# ryl.toml
[rules]
trailing-spaces = "enable"
new-line-at-end-of-file = "enable"

Run and branch on exit codes

ryl check <path|file ...>    # files and/or directories; `ryl check .` recurses, honouring .gitignore
ryl check --list-files .      # preview which files would be linted, then exit

Exit codes from a plain lint run (--fix/--diff differ, see below):

• 0: no errors (warning-level findings still exit 0 unless --strict).
• 1: error-level findings, invalid YAML, or an unreadable path.
• 2: usage error (no inputs, bad flags), no config / no rules enabled, or --strict with only warnings.

Add --strict to make warnings fail (exit 2); --no-warnings reports only errors.

Fix or preview

• --fix applies safe fixes in place, then reports remaining problems.
• --diff previews the fixes (unified diff on stdout), writes nothing, exits 1 if any file would change. Use --diff for a non-mutating check, --fix to apply.
• --diff only previews: files it can't represent as a text diff are skipped with a notice and don't change the exit code. --fix writes in place, but an unparsable file is reported (not rewritten) and still exits 1.

Machine-readable output

-f/--format: parsable or github (line-oriented, stderr) for parsing diagnostics; junit or gitlab (stdout) for CI report artifacts. --format is repeatable and each -o/--output-file binds to the preceding format, so console + a report file can be produced together:

ryl check --format github --format gitlab -o code-quality.json .

Configuration: YAML vs TOML

• YAML (.yamllint, YAMLLINT_CONFIG_FILE) is yamllint-compatible (extends: default/relaxed, presets). YAMLLINT_CONFIG_FILE takes a YAML config only: a .toml target errors (use -c/-d or project discovery for ryl TOML).
• TOML (ryl.toml, .ryl.toml, [tool.ryl]) holds ryl-only features: [files] globs, [markdown] embedding, [output] destinations, per-line ignores, ryl-only rules.

Migrate a yamllint setup with ryl --migrate-configs (add --migrate-write to apply).

Stdin, directives, Markdown

• - reads stdin; pair with --stdin-filename <PATH> so diagnostics, config discovery, and filtering behave as if the path were on disk.
• Suppress rules inline with # ryl disable/enable/disable-line, or a first-line # ryl disable-file.
• --markdown lints YAML front matter and fenced yaml/yml blocks in Markdown.

Full documentation

• Agent guide: https://ryl-docs.pages.dev/using-ryl-with-ai-agents/
• Quick start: https://ryl-docs.pages.dev/getting-started/quickstart/
• Migrating from yamllint: https://ryl-docs.pages.dev/getting-started/migrating-from-yamllint/
• Output formats: https://ryl-docs.pages.dev/output-formats/
• Rules: https://ryl-docs.pages.dev/rules/