Third leg of the distribution sweep (companion to #16 for MCP Registry, #17 for Glama). Tracks getting wyre-technology/msp-claude-plugins indexed on skillsmp.com — the largest public registry of agent skills (96K+ skills per recent Medium coverage).
How skillsmp works
No submission API, no form, no PR ingest. skillsmp is purely auto-crawled. From their docs and public coverage:
SkillsMP's scraper regularly syncs with GitHub to fetch the latest skills and updates. When a skill repository is updated on GitHub, the changes will appear on SkillsMP shortly after the next sync. Skills on the marketplace are sourced from public GitHub repositories, with low-quality repos filtered out (minimum 2 stars).
Discovery criteria (inferred from coverage):
- Public GitHub repo
- ≥2 stars
- Contains
SKILL.md files with valid YAML frontmatter (name, description required per the official spec)
API for verifying indexing: GET https://skillsmp.com/api/v1/skills/search?q=<query>.
Current state (2026-05-20)
wyre-technology/msp-claude-plugins: public ✓, 24 stars ✓, 267 SKILL.md files with valid frontmatter ✓ — yet 0 results in skillsmp for q=msp-claude-plugins.wyre-technology/claudeCISBenchmarks: indexed, multiple skills surface for q=wyre-technology. Proves skillsmp does crawl our org.
So we qualify on paper but aren't appearing.
Diagnosis
Comparing the indexed repo to the missing one:
| Repo |
Skill path |
Indexed? |
claudeCISBenchmarks |
skills/<name>/SKILL.md (2 levels deep) |
✅ |
msp-claude-plugins |
msp-claude-plugins/<vendor>/<vendor>/skills/<name>/SKILL.md (4-5 levels deep, double-nested vendor dir) |
❌ |
But depth alone isn't the blocker — other indexed skills sit 7+ levels deep (e.g. team-michael/hermes-agent/ignored/local/profiles/boris/skills/devops/msp-support-ticket). Likely causes, in order of probability:
- Scraper backlog. The repo is relatively new; the crawler may simply not have hit it yet.
- The double-nested
msp-claude-plugins/msp-claude-plugins/ folder — a confusing path that may trip the scraper's skill-detection heuristic.
- Frontmatter dialect. Our SKILL.md files include extension fields (
when_to_use, triggers) alongside the standard name/description. The scraper should tolerate this since both required fields are present, but worth confirming.
Proposed steps
- Wait two weeks first. If the scraper backlog is the cause, this resolves itself. Re-query the API to confirm.
- If still 0 results after two weeks: open a support ticket / email to skillsmp owner (the @punkpeye Discord, same person who runs Glama, per the homepage). Provide the repo URL and ask whether the structure is supported.
- If the structure is the blocker: flatten the double-nested path (rename
msp-claude-plugins/msp-claude-plugins/ to msp-claude-plugins/plugins/ or just plugins/) — cleaner anyway. Symlinks won't help: GitHub trees don't follow them.
- Once indexed: the daily stars-watcher digest could be extended with a
skillsmp coverage section, mirroring the registry/Glama pattern. Out of scope for this issue.
Out of scope
Acceptance
GET https://skillsmp.com/api/v1/skills/search?q=msp-claude-plugins returns a non-empty result set with our skills visible by name and author wyre-technology.
Third leg of the distribution sweep (companion to #16 for MCP Registry, #17 for Glama). Tracks getting
wyre-technology/msp-claude-pluginsindexed on skillsmp.com — the largest public registry of agent skills (96K+ skills per recent Medium coverage).How skillsmp works
No submission API, no form, no PR ingest. skillsmp is purely auto-crawled. From their docs and public coverage:
Discovery criteria (inferred from coverage):
SKILL.mdfiles with valid YAML frontmatter (name,descriptionrequired per the official spec)API for verifying indexing:
GET https://skillsmp.com/api/v1/skills/search?q=<query>.Current state (2026-05-20)
wyre-technology/msp-claude-plugins: public ✓, 24 stars ✓, 267 SKILL.md files with valid frontmatter ✓ — yet 0 results in skillsmp forq=msp-claude-plugins.wyre-technology/claudeCISBenchmarks: indexed, multiple skills surface forq=wyre-technology. Proves skillsmp does crawl our org.So we qualify on paper but aren't appearing.
Diagnosis
Comparing the indexed repo to the missing one:
claudeCISBenchmarksskills/<name>/SKILL.md(2 levels deep)msp-claude-pluginsmsp-claude-plugins/<vendor>/<vendor>/skills/<name>/SKILL.md(4-5 levels deep, double-nested vendor dir)But depth alone isn't the blocker — other indexed skills sit 7+ levels deep (e.g.
team-michael/hermes-agent/ignored/local/profiles/boris/skills/devops/msp-support-ticket). Likely causes, in order of probability:msp-claude-plugins/msp-claude-plugins/folder — a confusing path that may trip the scraper's skill-detection heuristic.when_to_use,triggers) alongside the standardname/description. The scraper should tolerate this since both required fields are present, but worth confirming.Proposed steps
msp-claude-plugins/msp-claude-plugins/tomsp-claude-plugins/plugins/or justplugins/) — cleaner anyway. Symlinks won't help: GitHub trees don't follow them.skillsmpcoverage section, mirroring the registry/Glama pattern. Out of scope for this issue.Out of scope
Acceptance
GET https://skillsmp.com/api/v1/skills/search?q=msp-claude-pluginsreturns a non-empty result set with our skills visible by name and authorwyre-technology.