doraval

Verify agent-context quality with the `dora` CLI. Whenever you author or edit a skill (SKILL.md), plugin, rule, or agent config, run dora to check it before reporting the work done. Also use when the user asks to check, diagnose, fix, or clean up agent context across Claude, Cursor, Codex, Copilot, or Grok. Not for ordinary source-code edits.

Use when testing scenario coverage in review tier 3.

Nested skill under packages/x/skills, used to test recursive discovery.

First skill in the multi-skill fixture repo, used to test fan-out validation.

Second skill in the multi-skill fixture repo, used to test fan-out validation.

Extracts examples for testing the agentskills.io spec validator. Use when validating a well-formed skill against the open spec.

Valid description, but the name field does not match the parent directory name.

Body deliberately exceeds the Level 2 line/token budget to exercise the progressive-disclosure warning.

Use doraval to validate, measure drift, and judge skills and plugins. Use when authoring or reviewing context engineering artifacts for AI coding agents (works for Codex too).

Use when you need a full-spec example with rich frontmatter, dynamic injection and substitutions.

A test skill for fixture purposes

Test

[unclosed list

Use when testing Windows CRLF line endings.

A skill without activation trigger phrases.

Use when testing doraval validate and drift commands.