MCP for AI assistants
Give Codex, Claude, Copilot, Gemini and Cursor reference access to SSTIM.
@sstim/mcp is a public stdio MCP server, read-only unless its operator enables public issue submission, published as io.github.w3c-cg/sstim in the official MCP Registry. It retrieves frozen released concepts, not live clinical findings. Your AI client provides the conversation, code search, comparisons and proposed changes.
Example 01
Install the MCP server in VS Code/Copilot
Make SSTIM terms available to your coding assistant with no SSTIM checkout.
- Ensure Node.js 20+ and npx are installed.
- Put the following JSON in your project’s .vscode/mcp.json, start the server from VS Code’s MCP command palette, then enable its tools in Copilot Agent.
{
"servers": {
"sstim": {
"type": "stdio",
"command": "npx",
"args": ["--yes", "@sstim/mcp@0.3.0"]
}
}
}Example 02
Install in Codex, Claude Code or Cursor
Reuse the same public package from another AI coding client.
- For Codex (CLI and VS Code extension), append this to ~/.codex/config.toml.
- Claude Code: use claude mcp add. Cursor: use .cursor/mcp.json with mcpServers instead of VS Code’s servers.
[mcp_servers.sstim]
command = "npx"
args = ["--yes", "@sstim/mcp@0.3.0"]Example 03
Ask which vocabulary terms match your software model
Map parameter fields onto shared ontology identifiers, flagging gaps rather than inventing names.
- Ask your agent to inspect an existing application schema or JSON document.
- Have it search SSTIM and then retrieve exact term records; review its proposed mapping yourself.
Prompt for your AI assistant
Inspect my JSON session model and use sstim_search_concepts and sstim_get_concept to propose mappings for signal, carrier frequency, modulation, channel, exposure and session fields. Return exact IRIs, release, mapping rationale and unresolved gaps. Do not invent SSTIM properties.
Example 04
Audit a scientific vocabulary in a paper
Replace ambiguous language with defined terms without misrepresenting mechanism or evidence.
- Ask the agent to search for each key term.
- Have it quote or paraphrase the definitions with release and provenance; verify whether distinctions like stimulus rate vs neural band are represented.
Prompt for your AI assistant
Find published SSTIM terms related to binaural stimulation, isochronous modulation and oscillatory bands. For each, give the canonical IRI, definition, relevant relationships and the source release. Distinguish asserted ontology facts from your interpretation.
Example 05
Compare two frozen releases
Detect whether term descriptions or mappings changed, useful for migrations and manuscript reproducibility.
- Use an exact IRI/CURIE and query the term in release 0.18.0 and 0.19.0.
- Compare the returned definitions and relationships. An absent term is a meaningful result, not an invitation to invent one.
Prompt for your AI assistant
Use sstim_get_concept to retrieve the same exact concept IRI from SSTIM v0.18.0 and v0.19.0. Compare labels, definitions, relationships, deprecation status and provenance. Report only verifiable differences.
Example 06
Crosswalk external research vocabularies
Inspect SKOS mappings and their strength when relating SSTIM to adjacent ontologies.
- Search for the relevant SSTIM concept.
- Inspect exactMatch, closeMatch, broadMatch and other predicate types rather than treating all mappings as identical.
Prompt for your AI assistant
Retrieve SSTIM terms related to sensory stimulation and list their external mappings with exact target IRIs, predicate types, and source provenance. Do not interpret a SKOS mapping as clinical validation.
Example 07
Create a reviewer-controlled contribution
Report a missing definition or propose an improvement without automatically modifying the ontology.
- Have the agent identify the exact term or confirm the gap.
- Call sstim_prepare_feedback to generate a reviewable Contribution Bridge link. The user chooses whether to open and submit an issue.
Prompt for your AI assistant
Find the SSTIM term most closely related to my proposed concept. Explain whether it is already covered. If not, prepare a Contribution Bridge link and a proposed definition for my review. Do not submit the issue or alter canonical files.
Example 08
Use SSTIM to guide a small data-pipeline refactor
Make a coding assistant check interoperability assumptions against the current reference before changing code.
- Ask the agent to inspect the repository schema and tests.
- Have it identify SSTIM term matches and gaps with source provenance, then propose a minimal version-pinned migration plan.
Prompt for your AI assistant
Inspect the stimulus-related fields in this repository. Search SSTIM MCP for canonical equivalents, propose a minimal version-pinned mapping plan, and add tests only for relationships supported by the published ontology. Ask before editing code.