Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@mcp-filesearch find all TypeScript files modified in the last week"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
The template every MCP server here is instantiated from: a stdio server with one registerTools seam, structured output, and annotation presets so a tool cannot be written without carrying its safety hints.
It is also the cheap place to retire SDK unknowns. The get_datetime example exists to exercise the SDK v2 surface end to end — handshake, annotations, structured output, schema rejection — so the servers that matter don't discover those on their own.
Tag | What it is |
| Proving ground — scaffold, example tool, tool-level tests |
| The backport — single subprocess seam, root/path validation, security test suite |
Still to land before v0.1.0 is tagged: the 1MCP wiring proof, scripts/instantiate.ts, the SDK-drift test and dependabot. The design notes live in Trilium.
Related MCP server: File Search Tool
Node ≥ 22. CI runs 22 and 24; .node-version pins 24 for local mise users.
npm ci
npm run build
npm testnpm test runs vitest against src/ through the SDK's own client over an in-memory transport, so no build and no spawned process are involved. npm run build is only needed for the bin entry and for npx over GitHub.
Use this template on GitHub — a fork would share history, which is not what you want.
Run the instantiation script. It substitutes __name__ and reports every replacement point it could not infer; the substitution contract lives in src/placeholders.ts. Until the script lands, substitute those points by hand.
Delete the example: src/tools/get-datetime.ts, its line in src/tools/index.ts, and its tests in test/.
Then add the first real tool — see AGENTS.md for the checklist.
"mcpbase": {
"command": "npx",
"args": ["-y", "github:GwylimWilliams/mcp-base"],
"tags": ["test"]
}The prepare script builds on install, which is what makes npx -y github:… work without a published package.
When a server takes scoped roots, they are passed as CLI args after the repo — never cwd, so the scoping holds whatever directory the gateway happens to run in.
CONVENTIONS.md — the canonical rule list
AGENTS.md — the same rules agent-facing, plus the per-tool checklist
docs/adding-a-tool.md — the worked walkthrough
SECURITY.md — the security posture and how to report a problem
MIT.
Current date and time in an IANA timezone, optionally shifted by whole days. The iso field is the wall clock in that zone, so its offset makes the instant unambiguous; unix is the same instant as epoch seconds, and human is a localized rendering.
| Name | Required | Description | Default |
|---|---|---|---|
| format | No | Adds a localized "human" string when set to "human". | iso |
| timezone | No | IANA timezone name, e.g. "UTC", "Europe/London", "Asia/Tokyo". | UTC |
| offsetDays | No | Whole days to shift the result, relative to now. |
| Name | Required | Description |
|---|---|---|
| iso | Yes | ISO 8601 wall-clock time in the requested zone, carrying that zone's UTC offset. |
| unix | Yes | Seconds since the Unix epoch. |
| human | No | Localized rendering; present only when format is "human". |
| timezone | Yes | Canonical IANA name of the resolved zone. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true and openWorldHint=false, so safety is covered. The description adds useful behavioral context by explaining that iso includes an offset that disambiguates the instant, that unix is the same instant as epoch seconds, and that human is a localized rendering.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, front-loaded with the core purpose and optional shift, followed by precise output-format semantics. Every clause contributes meaningful information with no filler.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Although an output schema exists, the description still usefully clarifies the meaning of the iso, unix, and human fields. Combined with the annotation safety hints and complete parameter schema, an agent has enough context to invoke the tool correctly; only explicit usage guidance is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so all three parameters are fully documented in the schema itself. The description restates the timezone and whole-day shift concepts but adds no syntax or format details beyond what the schema already provides, making the baseline 3 appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific resource and scope: current date and time, in an IANA timezone, optionally shifted by whole days. It is immediately clear what the tool returns and how its output is affected by parameters.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
There are no sibling tools and no alternatives to distinguish from, but the description does not explicitly state when to use this tool versus other time-related approaches. Usage is strongly implied by the resource and parameter set, which is adequate for a simple utility.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.0get_datetimeScored across 1 tool
The server exposes only one tool, so there is no possibility of confusing it with another tool. Its purpose is clearly stated as retrieving current date and time.
The single tool name get_datetime follows a clear snake_case verb_noun convention. There are no other names to conflict with or create inconsistency.
The server is named mcp-filesearch, but it provides only a datetime tool, which is unrelated to file search. A single trivial tool is an extreme mismatch for the apparent file-search purpose.
For a file-search server, there are no tools for searching, reading, listing, or retrieving files. The surface is severely incomplete and cannot support the stated domain.
Search and get install details on MCP servers, right from your agent -- a unified marketplace index.
Browse and manage files in your Moxt AI workspace from any MCP client.
Agent-driven search: build, import, tune, search, and score result quality — all over MCP.
