While auditing public Claude skills with SkillVerify — a description-vs-body verifier for SKILL.md files — I found that the JustOneAPI skill family published on clawhub.com, all of which appear to be produced from the same template that this repo backs, share a single recurring count-mismatch defect.
For each skill, the frontmatter.description advertises a specific number of accessible operations, but the SKILL.md body documents only 4 backticked tool/operation names. The mismatch ranges from off-by-3 (tiktok, kuaishou) to a 39-operation gap (douyin-xingtu).
Affected skills
| Clawhub skill |
Description claim |
Body documents |
Gap |
justoneapi/justoneapi-douyin-xingtu |
"across 43 operations" |
4 |
39 |
justoneapi/justoneapi-xiaohongshu-pgy |
"across 25 operations" |
4 |
21 |
justoneapi/justoneapi-imdb |
"19 operations" |
4 |
15 |
justoneapi/justoneapi-tiktok |
"7 operations" |
4 |
3 |
justoneapi/justoneapi-kuaishou |
"7 operations" |
4 |
3 |
justoneapi/justoneapi-douban |
"6 operations" |
4 |
2 |
Why this is a real defect
This is a semantic description-vs-body mismatch — both halves of each SKILL.md are valid Markdown (so a structural lint cannot catch it), but the user-facing claim contradicts the documented surface. A user who reads "43 operations" expects to find 43 callable operations and finds 4.
The defect class matches the same family of bugs documented in similar audits — e.g., manojbajaj95/canva-2 (22 tools claimed, 18 documented), rnwy/rnwy (8 tools claimed, 7 listed).
Suggested fix
Two options, depending on whether the body is intentionally a 4-operation sample or whether documentation is incomplete:
- (a) Update each skill's description count to match the actual documented operation surface (e.g., "4 sample operations" instead of "43 operations") — keeps the skill metadata honest.
- (b) Add the missing operations to each skill's body so the body matches the description count.
Or update the upstream generator that produces these skills' descriptions (likely something that pulls a count from the API spec while the body is hand-curated to a small sample).
While auditing public Claude skills with SkillVerify — a description-vs-body verifier for
SKILL.mdfiles — I found that the JustOneAPI skill family published on clawhub.com, all of which appear to be produced from the same template that this repo backs, share a single recurring count-mismatch defect.For each skill, the
frontmatter.descriptionadvertises a specific number of accessible operations, but theSKILL.mdbody documents only 4 backticked tool/operation names. The mismatch ranges from off-by-3 (tiktok,kuaishou) to a 39-operation gap (douyin-xingtu).Affected skills
justoneapi/justoneapi-douyin-xingtujustoneapi/justoneapi-xiaohongshu-pgyjustoneapi/justoneapi-imdbjustoneapi/justoneapi-tiktokjustoneapi/justoneapi-kuaishoujustoneapi/justoneapi-doubanWhy this is a real defect
This is a semantic description-vs-body mismatch — both halves of each
SKILL.mdare valid Markdown (so a structural lint cannot catch it), but the user-facing claim contradicts the documented surface. A user who reads "43 operations" expects to find 43 callable operations and finds 4.The defect class matches the same family of bugs documented in similar audits — e.g.,
manojbajaj95/canva-2(22 tools claimed, 18 documented),rnwy/rnwy(8 tools claimed, 7 listed).Suggested fix
Two options, depending on whether the body is intentionally a 4-operation sample or whether documentation is incomplete:
Or update the upstream generator that produces these skills' descriptions (likely something that pulls a count from the API spec while the body is hand-curated to a small sample).