I ran the checklist from the skill authoring guide across the 64 SKILL.md files under plugins/. Three things came out of it. One is a documented validation rule, two are guidance. None of them produced a failure I could reproduce, so this is a list to take or leave rather than a bug report.
name: anthropic-frontend-design uses a reserved word
plugins/frontend-design-skills/skills/anthropic-frontend-design/SKILL.md, line 2.
The frontmatter reference in that guide says the name field "Cannot contain reserved words: anthropic, claude", and the naming section lists anthropic-helper and claude-tools under Avoid. I did not get a hard load error out of this, so what I can claim is that the name conflicts with a documented restriction. I cannot claim it is broken.
The prefix earns its place, since it pairs with openai-frontend-design in the same plugin and tells a reader where the guidance came from. Renaming changes the name Claude invokes the skill by, and it also reaches the README table at line 377 and the release asset anthropic-frontend-design.zip. That is a breaking change for anyone who scripted the download, so the call is yours and leaving it alone is a defensible answer.
polar-billing SKILL.md body runs to 815 lines
plugins/polar-skills/skills/polar-billing/SKILL.md is 820 lines with the frontmatter closing on line 5. The guide asks for a body under 500 lines, because everything past the frontmatter loads the moment the skill triggers.
The repo already has the pattern that fixes this. mongodb-schema-design/SKILL.md is a short index that points at fourteen files in references/, each one link deep.
26 reference files over 100 lines with no contents list
The guide asks for a table of contents at the top of any reference file longer than 100 lines, so that a partial read still shows the full scope of what is in the file. Twenty-six files pass that line count without one.
By plugin: mongodb-skills 17, react-skills 4, polar-skills 3, python-skills 1, agent-browser 1.
Full list, lines then path
457 plugins/mongodb-skills/skills/mongodb-atlas-stream-processing/references/pipeline-patterns.md
418 plugins/mongodb-skills/skills/mongodb-schema-design/references/fundamental-embed-vs-reference.md
312 plugins/react-skills/skills/react-view-transitions/references/patterns.md
310 plugins/python-skills/skills/python-guidelines/references/idiomatic-patterns.md
308 plugins/react-skills/skills/react-view-transitions/references/css-recipes.md
304 plugins/mongodb-skills/skills/mongodb-atlas-stream-processing/references/development-workflow.md
298 plugins/mongodb-skills/skills/mongodb-atlas-stream-processing/references/connection-configs.md
273 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-schema-versioning.md
256 plugins/polar-skills/skills/polar-local-environment/rules/troubleshooting.md
226 plugins/mongodb-skills/skills/mongodb-schema-design/references/fundamental-document-size.md
222 plugins/react-skills/skills/react-view-transitions/references/nextjs.md
190 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-time-series-collections.md
184 plugins/react-skills/skills/react-view-transitions/references/implementation.md
178 plugins/mongodb-skills/skills/mongodb-atlas-stream-processing/references/sizing-and-parallelism.md
167 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-polymorphic.md
166 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-document-versioning.md
162 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-computed.md
159 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-outlier.md
150 plugins/mongodb-skills/skills/mongodb-atlas-stream-processing/references/output-diagnostics.md
143 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-archive.md
131 plugins/mongodb-skills/skills/mongodb-schema-design/references/fundamental-schema-validation.md
123 plugins/mongodb-skills/skills/mongodb-schema-design/references/source-query-stats.md
120 plugins/polar-skills/skills/polar-local-environment/rules/payment-testing.md
118 plugins/agent-browser/skills/agent-browser/references/webgpu.md
108 plugins/polar-skills/skills/polar-local-environment/rules/view-logs.md
102 plugins/mongodb-skills/skills/mongodb-schema-design/references/pattern-bucket.md
Several of these are close enough to the threshold that a contents list would be longer than the section it indexes. pipeline-patterns.md at 457 lines and fundamental-embed-vs-reference.md at 418 are the two where it would actually change what Claude sees on a head -100 preview.
One finding I dropped
The same check first flagged agent-browser/references/video-recording.md for linking to commands.md, on the reasoning that this puts commands.md two levels from SKILL.md. That reading is wrong. SKILL.md names references/commands.md on lines 462 and 495, so the file is one level deep and the guidance is satisfied. I corrected my own check instead of filing that one.
I opened each of the files above and read them before writing this, and I am happy to be told any of it is not worth doing. What brought me here is a checker I maintain against the published skill guidance, so I am saying that up front rather than leaving you to wonder. I have not run these skills through Claude to observe a discovery or context failure, so every claim here rests on the guide's wording and on the line counts. If you would rather not get this kind of issue, say so and I will not send another.
I ran the checklist from the skill authoring guide across the 64
SKILL.mdfiles underplugins/. Three things came out of it. One is a documented validation rule, two are guidance. None of them produced a failure I could reproduce, so this is a list to take or leave rather than a bug report.name: anthropic-frontend-designuses a reserved wordplugins/frontend-design-skills/skills/anthropic-frontend-design/SKILL.md, line 2.The frontmatter reference in that guide says the
namefield "Cannot contain reserved words:anthropic,claude", and the naming section listsanthropic-helperandclaude-toolsunder Avoid. I did not get a hard load error out of this, so what I can claim is that the name conflicts with a documented restriction. I cannot claim it is broken.The prefix earns its place, since it pairs with
openai-frontend-designin the same plugin and tells a reader where the guidance came from. Renaming changes the name Claude invokes the skill by, and it also reaches the README table at line 377 and the release assetanthropic-frontend-design.zip. That is a breaking change for anyone who scripted the download, so the call is yours and leaving it alone is a defensible answer.polar-billingSKILL.md body runs to 815 linesplugins/polar-skills/skills/polar-billing/SKILL.mdis 820 lines with the frontmatter closing on line 5. The guide asks for a body under 500 lines, because everything past the frontmatter loads the moment the skill triggers.The repo already has the pattern that fixes this.
mongodb-schema-design/SKILL.mdis a short index that points at fourteen files inreferences/, each one link deep.26 reference files over 100 lines with no contents list
The guide asks for a table of contents at the top of any reference file longer than 100 lines, so that a partial read still shows the full scope of what is in the file. Twenty-six files pass that line count without one.
By plugin: mongodb-skills 17, react-skills 4, polar-skills 3, python-skills 1, agent-browser 1.
Full list, lines then path
Several of these are close enough to the threshold that a contents list would be longer than the section it indexes.
pipeline-patterns.mdat 457 lines andfundamental-embed-vs-reference.mdat 418 are the two where it would actually change what Claude sees on ahead -100preview.One finding I dropped
The same check first flagged
agent-browser/references/video-recording.mdfor linking tocommands.md, on the reasoning that this putscommands.mdtwo levels from SKILL.md. That reading is wrong.SKILL.mdnamesreferences/commands.mdon lines 462 and 495, so the file is one level deep and the guidance is satisfied. I corrected my own check instead of filing that one.I opened each of the files above and read them before writing this, and I am happy to be told any of it is not worth doing. What brought me here is a checker I maintain against the published skill guidance, so I am saying that up front rather than leaving you to wonder. I have not run these skills through Claude to observe a discovery or context failure, so every claim here rests on the guide's wording and on the line counts. If you would rather not get this kind of issue, say so and I will not send another.