| name | skill-name-here |
| description | One or two sentences. Lead with what it does, then the trigger: 'Use when …' with concrete keywords the model will match on. Keep under 1024 characters. |
| version | 0.1.0 |
| domain | research |
| tags | ["tag-one","tag-two"] |
Skill Name Here
One-line statement of the transformation this skill performs: input → output.
Purpose
What this skill is for, in two or three sentences. What problem it removes, and
who benefits. Grounded and concrete — no claims you can't demonstrate.
When it fires
- Keyword triggers:
keyword, another keyword
- File triggers:
*.ext, path/pattern/**
- Command triggers:
/some-command
Inputs
| Input | Required | Notes |
|---|
source | yes | What the user supplies (URL, file, topic). |
options | no | Tunable parameters and their defaults. |
Workflow
- Step one — what to do, and the check that proves it worked.
- Step two — …
- Step three — …
Keep this section as a checkable procedure, not prose. Each step should have an
observable result so the agent can verify before moving on.
Output contract
Describe exactly what the skill produces. If it emits structured output, show the
shape (matches manifest.json → outputs):
{
"field": "type — meaning"
}
Tools & MCP
- Tools used:
Read, WebFetch, …
- MCP dependencies (if any): name them and what they provide. List the same set in
manifest.json → mcp_dependencies.
Quality bar
Example
See examples/. Show one real input and the resulting output.
Built on SIP — Starlight Intelligence Protocol
Substrate: starlightintelligence.org/protocol v1.1.0
Layers used: [file-contract, attestation, sovereignty]
Vertical: starlight-agent-skills · portable capability layer
