See it in action
One of 11 test cases the skill fixed. Same model, same prompt, with and without the skill.
PromptWe want to install the Coss UI Tabs component using the recommended CLI command. What is the exact command?
What was checked, without and with the skill| Checked | Without skill | With skill |
|---|
| The installation command uses 'npx shadcn@latest add @coss/tabs'. | Missed | Passed |
|---|
What's inside
The instructions your model reads when the skill loads.
SKILL.md107 lines4 files
coss ui
coss ui is a component library built on Base UI with a shadcn-like developer experience plus a large particle catalog.
What this skill is for
Use this skill to:
- pick the right coss primitive(s) for a UI task
- write correct coss usage code (imports, composition, props)
- avoid common migration mistakes from shadcn/Radix assumptions
- reference particle examples to produce practical, production-like patterns
Source of truth
- coss components docs:
apps/ui/content/docs/components/*.mdx
https://github.com/cosscom/coss/tree/main/apps/ui/content/docs/components
- coss particle examples:
apps/ui/registry/default/particles/p-*.tsx
https://github.com/cosscom/coss/tree/main/apps/ui/registry/default/particles
- coss particles catalog:
https://coss.com/ui/particles
- docs map for agents:
https://coss.com/ui/llms.txt
Out of scope
- Maintaining coss monorepo internals/build pipelines.
- Editing registry internals unless explicitly requested.
Principles for agent output
- Use existing primitives and particles first before inventing custom markup.
- Prefer composition over custom behavior reimplementation.
- Follow coss naming and APIs from docs exactly.
- Keep examples accessible and production-realistic.
- Prefer concise code that mirrors coss docs/particles conventions.
- Assume Tailwind CSS v4 conventions in coss examples and setup guidance.
Critical usage rules
Always apply before returning coss code:
Track record
Benchmark
16 / 21
test cases passed, vs 8 / 21 without
Adoption
- Users
- 0
- Times used
- 0
Safety and source details
SkillSafety
✓ 3 of 3 passed✓Code scan15 of 15 checks clean
0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026
View 15 scanner checks
live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing
✓IntentSafe
LLM judge · does the skill do what it claims
✓ContentClean
LLM judge · policy & harmful-content review
Source
ImportedLicense MITImported 8/3/2026Version v1Bundled 3 files