Exit Codes

Exit codes

Code	Meaning
0	Run succeeded; no error-severity diagnostics, no --max-warnings overflow.
1	Run produced at least one error, or warnings exceeded --max-warnings=N.
2	Usage, config, or I/O problem (--format=bogus, unreadable config, etc.).

Suppressing diagnostics

Three independent mechanisms, each documented in DESIGN-0001 §Suppression model.

In-source markers (Markdown kinds only)

CLAUDE.md, SKILL.md, command, and agent files accept HTML-comment markers. The marker silences diagnostics on its own line and the next non-blank line.

<!-- claudelint:ignore=skills/require-name -->              <!-- same line or next non-blank line -->
<!-- claudelint:ignore=skills/require-name,style/no-emoji --> <!-- multiple IDs -->

<!-- claudelint:ignore-file=skills/require-name -->        <!-- whole-file -->

JSON-backed artifacts (hooks.json, plugin manifests) do not recognize in-source markers — JSON has no standard comment syntax. Use config-level suppression for those kinds.

Config-level rule toggles

In .claudelint.hcl, disable a rule for every artifact or scope it to a subset of paths:

rule "skills/require-name" {
  enabled = false
}

rule "style/no-emoji" {
  # Only silence this rule inside vendored docs.
  paths = ["vendor/**/*.md"]
}

ignore {
  paths = [
    "testdata/**",
    "vendor/**",
  ]
}

Unknown-ID safety net

A typo in a rule "<id>" block does not silently disable the real rule. claudelint emits a meta/unknown-rule warning pointing at the offending config so the typo is visible.