It lets MCP clients query a UniFi network controller through a read-only, spec-driven interface.
Discover entities: unifi_list_entities lists available OpenAPI tags with read/write op counts.
Inspect an entity: unifi_describe_entity shows operations, parameters, and fields.
Run read queries: unifi_get invokes GET operations with path/query parameters.
Write operations are gated: unifi_invoke exists but is disabled unless UNIFI_ALLOW_WRITES=true.
Dynamic spec resolution: loads the OpenAPI spec from cache, live gateway, stale cache, or bundled fallback.
No per-resource code: new UniFi entities/operations become available when the spec is updated.
Configurable access: supports CA cert pinning, opt-in insecure TLS, timeouts, cache controls, and log levels.
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., "@unifi-mcp show me the connected clients on my UniFi network"
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.
A Model Context Protocol server for the UniFi Local Network Integration API. Query your UniFi network from Claude Desktop, Claude Code, Cursor, or any MCP client.
Phase 1 — read-only, four generic tools, spec-driven interface generated at runtime from the UniFi OpenAPI spec.
Related MCP server: UniFi MCP Server
The server fetches the OpenAPI spec from your gateway at startup (or uses a bundled fallback) and derives all tool behaviour from it — there is no per-resource code. Adding a new UniFi entity or operation requires no code change; it becomes available as soon as the updated spec is fetched.
The four tools follow a discover-then-query pattern:
Tool | Purpose |
| List available entities (derived from OpenAPI tags) |
| Operations, parameters, and fields for one entity |
| Invoke a read (GET) operation on an entity |
| Invoke a write operation — disabled until write support ships |
The agent calls unifi_list_entities to discover what is available, then
unifi_describe_entity on an entity to learn its operations and field names,
then unifi_get to retrieve data.
unifi_invoke is defined and registered but gated: it returns an error unless
UNIFI_ALLOW_WRITES=true is set. The default is read-only. Write support will
ship in a later phase.
The server resolves the OpenAPI spec in order:
Fresh local cache (age < UNIFI_SPEC_FRESHNESS_MS, default 24 h)
Live fetch from the gateway (UNIFI_BASE_URL/proxy/network/integration/v1)
Stale local cache (if the live fetch fails)
Bundled spec (a static-inference draft, shipped with the package as a last-resort fallback)
Run pnpm update-spec to update the bundled spec from a live gateway.
Copy .env.example to .env and fill in the required values.
Variable | Required | Default | Description |
| yes | — | Gateway address, e.g. |
| yes | — | Integration API key (see below) |
| no | — | Path to the controller's CA certificate (PEM) |
| no |
| Disable TLS verification — last resort only |
| no |
| Enable write operations via |
| no |
| Per-request timeout in milliseconds |
| no | — | Override the OpenAPI spec URL fetched from the gateway |
| no | — | Use a local spec file as the bundled fallback |
| no |
| Where the cached spec is written |
| no |
| Max age of the cached spec in milliseconds |
| no |
| Pino log level ( |
In the UniFi Network application: Settings → Integrations → Add Integration.
Copy the generated key into UNIFI_API_KEY.
UniFi gateways use self-signed certificates. The recommended approach is to pin the controller's CA certificate:
UNIFI_CA_CERT=/path/to/controller-ca.pemExport the certificate from the UniFi console or your browser and provide the path above. This keeps TLS verification enabled.
UNIFI_INSECURE_TLS=true disables certificate verification entirely. Use it
only as a last resort — it exposes connections to man-in-the-middle attacks.
The server prints a warning to stderr on startup when it is set.
pnpm install
pnpm build
node dist/cli.js # or: unifi-mcp (after npm install -g @robinbowes/unifi-mcp)Note: The package is published as @robinbowes/unifi-mcp (scoped), but the CLI
command is unifi-mcp — unchanged.
The transport is stdio: the client spawns the binary and speaks MCP over its
stdin/stdout. The server reads configuration from the process environment only
— it does not load .env — so the gateway address and API key must come from
the client's own config block.
claude mcp add unifi -s user \
-e UNIFI_BASE_URL=https://192.168.1.1 \
-e UNIFI_API_KEY=your-integration-api-key \
-- npx -y @robinbowes/unifi-mcp-s user makes the server available in every project; -s local (the default)
limits it to the current one. Avoid -s project with a literal key — that scope
writes to .mcp.json, which is committed. Confirm with claude mcp list.
A global install drops the npx resolution from every launch:
npm install -g @robinbowes/unifi-mcp
claude mcp add unifi -s user \
-e UNIFI_BASE_URL=https://192.168.1.1 \
-e UNIFI_API_KEY=your-integration-api-key \
-- unifi-mcpAdd an entry to claude_desktop_config.json and restart the app. It lives in
~/Library/Application Support/Claude/ on macOS and %APPDATA%\Claude\ on
Windows.
{
"mcpServers": {
"unifi": {
"command": "/usr/local/bin/node",
"args": ["/usr/local/lib/node_modules/@robinbowes/unifi-mcp/dist/cli.js"],
"env": {
"UNIFI_BASE_URL": "https://192.168.1.1",
"UNIFI_API_KEY": "your-integration-api-key"
}
}
}
}If the file already exists, merge the unifi key into the existing
mcpServers object — pasting the block over the whole file drops any other
servers you have configured.
Use absolute paths. Claude Desktop is a GUI application and does not inherit a
login shell's PATH, so a bare node, npx, or unifi-mcp fails to spawn.
command -v node and npm root -g give the paths for your machine.
Other stdio clients take the same command/args/env shape — Cursor and
Windsurf under mcpServers, Zed under context_servers.
pnpm install && pnpm build
claude mcp add unifi-dev \
-e UNIFI_BASE_URL=https://192.168.1.1 \
-e UNIFI_API_KEY=your-integration-api-key \
-- node /absolute/path/to/unifi-mcp/dist/cli.jsSymptom | Cause |
Server exits at once with |
|
| Relative command in a GUI client; use an absolute path |
Tools load, every query fails on TLS | Self-signed gateway certificate; set |
Entity list looks unfamiliar or over-large | Gateway was unreachable at startup and the bundled draft spec was used as fallback |
For the last two, set UNIFI_LOG_LEVEL=info and read the client's server log:
the server logs spec resolved with the source it used, and warns explicitly
when it falls back to the bundled spec.
pnpm install
task hooks-install # install the git hooks (once per clone, and after they change)
pnpm dev # run from source with stdio transport
pnpm test # unit + component tests
pnpm verify # format + lint + typecheck + test
pnpm update-spec # refresh the bundled spec from a live gateway
pnpm smoke # build, then exercise the tools against a live controller (.env)task hooks-install is not optional bookkeeping: besides the pre-commit and
commit-msg checks, the hooks re-run pnpm install --frozen-lockfile after a
checkout, merge or rebase, which is what keeps node_modules matching the
lockfile for tools that read it directly (tsserver, ALE, node dist/cli.js).
pnpm smoke builds the server and drives it as a real MCP client over stdio
against the controller configured in your .env. It fetches the live spec,
lists and describes entities, runs a couple of read queries, and confirms the
read-only gate refuses a write. It is read-only: nothing on the network is
changed, and the API key is never printed.
MIT.
Describe one entity: its operations, path/query parameters, and whether each is read (GET) or write.
| Name | Required | Description | Default |
|---|---|---|---|
| entity | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description must fully disclose behavior. It mentions the output includes read/write classification, but omits specifics like side effects (likely none), auth requirements, error handling, or data boundaries.
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?
Single concise sentence with no redundancy. All content is relevant and front-loaded.
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?
Given no output schema, no annotations, and simple input, the description fails to cover expected output format, valid entity values, or usage prerequisites. Agents may need to infer or experiment.
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 coverage is 0% and the description does not clarify the 'entity' parameter (e.g., if it's a name, ID, or endpoint path). No validation or enumeration is provided, leaving agents guessing.
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 clearly states the tool's function: describing an entity's operations, parameters, and HTTP method. This distinguishes it from siblings like unifi_list_entities (list all) and unifi_get (general get) by focusing on structural description.
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?
The description implies use when needing to understand an entity's API structure, but lacks explicit guidance on when not to use it or alternatives. Siblings are provided but not contrasted.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Invoke a read (GET) operation for an entity. Provide pathParams/query as needed.
| Name | Required | Description | Default |
|---|---|---|---|
| query | No | ||
| entity | Yes | ||
| pathParams | No | ||
| operationId | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden. It discloses that the operation is a read (GET), implying non-destructive behavior, but omits details about authentication, error handling, rate limits, or side effects. The minimal description fails to adequately inform the agent.
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?
The description is extremely concise (one sentence), but it lacks structure and front-loads only a generic purpose. Every word is used, but the sentence does not deliver comprehensive information. It is adequate in length but not well-organized.
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?
Given the absence of output schema and annotations, the description is far from complete. It does not describe return values, error states, or parameter constraints. The tool has moderate complexity (4 parameters, including nested objects), but the description provides insufficient context for correct usage.
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 0%, meaning the description adds no meaning beyond the schema. While it mentions 'pathParams/query as needed', it does not explain the required parameters 'entity' and 'operationId', nor provide context for their values. This leaves the agent without sufficient understanding of parameter usage.
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 it invokes a read (GET) operation for an entity, providing a general purpose. However, it does not clearly differentiate from sibling tools like unifi_list_entities or unifi_describe_entity, which might also involve reading. The phrase 'read (GET) operation' is moderately clear but lacks specificity.
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?
No guidance on when to use this tool versus alternatives. The description does not mention scenarios, prerequisites, or exclusions. Sibling tools exist but no comparison is provided, leaving the agent to guess the appropriate context.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Invoke any operation by id (including writes). Gated off unless UNIFI_ALLOW_WRITES=true. Post-v1 write path.
| Name | Required | Description | Default |
|---|---|---|---|
| body | No | ||
| query | No | ||
| entity | Yes | ||
| pathParams | No | ||
| operationId | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Discloses that the tool can perform writes and is gated by an environment variable, which is useful. However, without annotations, it omits details on side effects, auth requirements beyond the env variable, rate limits, or idempotency, leaving significant behavioral gaps.
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 concise sentences, each adding distinct value: the core functionality and an important constraint. No redundancy or unnecessary words.
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?
Given 5 parameters, no output schema, and no annotations, the description is too sparse to fully guide the agent. It lacks details on valid entities/operations, request formatting, and what the tool returns, making it incomplete for reliable invocation.
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?
The description implies that operationId is the operation identifier from 'by id', but provides no explanation for entity, body, query, or pathParams. With 0% schema description coverage, the description fails to add meaning beyond the schema, making parameter usage unclear.
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 clearly states 'Invoke any operation by id (including writes)' which specifies the action (invoke) and the resource (operation by id). It distinguishes from siblings which are for listing, describing, and getting entities, not invoking operations.
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?
Provides context: 'Gated off unless UNIFI_ALLOW_WRITES=true' and 'Post-v1 write path', but lacks explicit guidance on when to use vs siblings (e.g., for creating/updating/deleting) and no when-not-to-use or alternative recommendations.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List UniFi entities (OpenAPI tags) exposed by this controller, with read/write op counts. Call first.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description must carry the behavioral burden. It discloses the operation (list) and the data (entities with op counts) but omits details like auth requirements, caching, or pagination. While the behavior is simple, additional transparency would be beneficial.
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?
The description is a single sentence of 120 characters, highly concise and front-loaded with the core action. Every word adds value, and there is no redundancy.
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?
For a simple list tool with no output schema, the description adequately covers purpose and usage hint ('Call first'). It is missing explicit statements about output format, but the simplicity makes it reasonably complete.
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?
There are no parameters, and the schema coverage is trivially 100%. The description correctly adds no parameter information, which is appropriate. Baseline for 0 parameters is 4.
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 clearly states the verb 'List' and the resource 'UniFi entities (OpenAPI tags)', and specifies it includes read/write op counts. It effectively distinguishes itself from sibling tools by implying it is a discovery tool.
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?
The phrase 'Call first' provides explicit usage guidance, indicating this tool should be invoked before others. However, it does not explicitly state when not to use or contrast with siblings, leaving some room for ambiguity.
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.
v0.2.0unifi_describe_entityunifi_getunifi_invokeunifi_list_entitiesScored across 4 tools
Each tool has a distinct purpose: list entities, describe one entity, perform read-only GET, and invoke any operation (including writes). There is slight overlap between unifi_get and unifi_invoke for reads, but descriptions clarify safety, making ambiguity low.
All tools follow the pattern unifi_<verb>, with verbs list, describe, get, invoke. The naming is predictable and consistent, with no mixing of styles.
Four tools is appropriate for a dynamic API wrapper: discovery (list, describe) and execution (get, invoke). The number is well-scoped and each tool earns its place.
The tool set covers full lifecycle of API interaction: list available endpoints, describe details, perform safe reads, and invoke any operation (including writes if enabled). No obvious gaps for its purpose.
Query your org's data in natural language — read-only MCP access to SQL, NoSQL, files & warehouses.
Official Microsoft MCP Server to query Microsoft Entra data using natural language
Query your warehouse or a CSV with Claude/ChatGPT over MCP, governed by table-level ACL + audit.
Let AI agents query data and act across all your business apps via MCP.
