What this page checks, in one pass
Paste a SKILL.md or AGENTS.md file above. If the file opens with a YAML frontmatter block (--- on its own line, a block of key: value pairs, then a closing ---), it is checked as a SKILL.md file: name present, 64 characters or fewer, lowercase letters/numbers/hyphens only, and free of the reserved words "anthropic" and "claude"; description present, non-empty, 1,024 characters or fewer, and written in third person; a body of roughly 500 lines or fewer; and forward-slash paths in any referenced files. These limits come from Anthropic's Skill authoring best-practices documentation, which Kimi Code CLI's own skill-creator guide mirrors almost exactly.
If no frontmatter block is found, the file is checked as a plain instructions file in the style of AGENTS.md — the open convention formalized by OpenAI, Google, Cursor, and Factory and since donated to the Linux Foundation's Agentic AI Foundation. That spec defines no required fields, so this mode only reports word/line counts and whether the file uses Markdown headings, rather than failing it against SKILL.md's frontmatter rules.
What it cannot tell you
Passing every check here does not mean an agent will select, load, or correctly follow the file. Skill discovery depends on how well the description matches a real user request, and instruction-following depends on the specific host and model. This page also cannot follow file references to verify that a SKILL.md keeps its linked reference files one level deep, as Anthropic's guidance recommends — check that manually by opening each linked file.
Treat a clean result as "the file is structurally valid," not "the file works." Anthropic's own recommended process is to test a Skill on a representative task with the models you plan to use before shipping it — see the resource-compatibility notes for how Veomark labels publisher-documented support versus our own testing.
Where these rules come from
SKILL.md name and description limits are documented at Anthropic's Skill authoring best practices (platform.claude.com) and repeated in Kimi Code CLI's bundled skill-creator skill, which is why this checker applies them to both hosts. The AGENTS.md fallback mode reflects the open agents.md specification rather than any single vendor's rules. Neither ruleset is enforced by every host that reads these files — Grok Build, for example, documents that it accepts extra frontmatter fields like license and compatibility without validating them. When in doubt, check the current documentation for the specific host you are targeting; the skills listed in the resource directory link to that documentation per platform.
Frequently Asked Questions (FAQ)
What does this actually check?
For a file with YAML frontmatter, it checks the SKILL.md rules published by Anthropic's Skill authoring docs and mirrored in Kimi Code CLI's own skill-creator guide: a name field of 64 characters or fewer using only lowercase letters, numbers, and hyphens (with "anthropic" and "claude" reserved), a non-empty description field of 1,024 characters or fewer written in third person, a body kept under roughly 500 lines, and forward-slash file paths instead of Windows-style backslashes. For a file with no frontmatter, it treats the file as a plain AGENTS.md-style instructions file and checks length and headings only, since the AGENTS.md specification defines no required fields.
Can this tell me whether my skill actually works?
No. It only checks the static structure of the file. Whether an agent selects the skill, reads the right reference files, and follows the instructions correctly depends on the host, the model, and the task — none of which this page can observe. Test the skill with a real task in the target tool before relying on it.
Why are "anthropic" and "claude" flagged as reserved words?
Anthropic's Skill authoring documentation states the name field cannot contain the words "anthropic" or "claude." This is a rule of Anthropic's own Skill platform, not a universal SKILL.md constraint — a name aimed only at Kimi Code CLI or Grok Build would not need to avoid those words, but avoiding them keeps a skill portable across hosts that do enforce it.
Is my SKILL.md or AGENTS.md file uploaded anywhere?
No. Parsing and checking run in this browser tab. Nothing you paste here is sent to a server, which matters if the file references internal tools, paths, or credentials.
Why does AGENTS.md mode check different things than SKILL.md mode?
AGENTS.md is a free-form Markdown convention with no required frontmatter or fixed schema, while SKILL.md is a specific format with required name and description fields. Running SKILL.md's frontmatter rules against a plain AGENTS.md file would produce false failures, so this page only applies the frontmatter checks when it detects a YAML block.
