splunk-dashboard-mcp
This server lets an MCP client interact with Splunk as the logged-in user (via SSO cookies) to read, create, and update dashboards, reports, and alerts, run SPL searches, and consult official docs.
Authentication & session:
splunk_login(browser SSO),whoamito see the acting user, and a keep-alive loop to keep the session warm.Dashboards:
list_dashboards,get_dashboard,check_dashboard_access,create_dashboard,update_dashboard— supports Dashboard Studio (JSON) and Classic (Simple XML); writes are access-checked and created under your profile.Reports:
list_reports,get_report,create_report,update_report— saved searches, optionally scheduled via cron.Alerts:
list_alerts,get_alert,create_alert,update_alert— scheduled saved searches with trigger conditions and actions.Search:
query_splunkruns SPL searches and returns results, with time range and result limits.Documentation-first guardrail:
splunk_docsreturns official doc references; write tools requiredocsConsulted: truebefore creating/updating anything.No deletion: intentionally no delete tools for dashboards, reports, or alerts.
Provides tools for interacting with Splunk's management REST API to list, read, create, and update dashboards (Dashboard Studio and Classic), run SPL searches, and manage reports and alerts with access control enforcement.
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., "@splunk-dashboard-mcp Show me the dashboards in the main app."
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.
splunk-platform-mcp
A Model Context Protocol (MCP) server for reading and building Splunk dashboards.
It talks to the Splunk management REST API (data/ui/views) as you — via browser
SSO cookies — so Splunk's own permissions decide what you can see
and edit. It supports both Dashboard Studio (JSON) and Classic (Simple XML)
dashboards, checks your edit access before updating, and always creates new
dashboards under your own profile.
It also enforces a documentation-first workflow: the write tools require you to confirm you consulted the official Splunk docs before authoring SPL or a dashboard definition.
How it works
Auth: Complete SSO once in a browser (
splunk_login); cookies are saved and reused for REST calls. Cookies carry your identity, so all reads/writes are ACL-enforced by Splunk. (Auth is SSO-cookie only — there is no token mode.)Reach: REST is reached through the web tier proxy path (
{base}/{locale}/splunkd/__raw/..., SSO-cookie friendly) with a fallback to the management port (:8089). The server probes which works and caches it. Pin it withSPLUNK_REST_MODE=proxy|mgmt.Read:
list_dashboards,get_dashboardreturn only what your account can see.Access check:
check_dashboard_accessreads the ACL and reportscan_write.Write:
create_dashboardcreates under your username (private by default);update_dashboardre-checkscan_writeand refuses if Splunk denies write.
Related MCP server: Tableau MCP Server
Tools
Tool | Purpose |
| Browser SSO; save cookies for REST. |
| Return official Splunk doc references (topic: spl/studio/classic/report/alert/rest). Use before authoring. |
| Show the Splunk user the session runs as (create owner). |
| Run an SPL search and return results (runs as you). |
| List dashboards you can see; filter by app/owner/name. |
| Full definition + ACL + detected format for one dashboard. |
| Report whether you can edit a dashboard before trying. |
| Create a new dashboard under your profile (Studio or Classic). Requires |
| Update an existing dashboard; refuses if |
| List / read reports (saved searches). |
| Create under your profile / update (access-checked). Requires |
| List / read alerts (scheduled triggering saved searches). |
| Create under your profile / update (access-checked). Requires |
There is intentionally no delete tool — this server does not delete dashboards, reports, or alerts.
Reports and alerts
Both are Splunk saved/searches objects. A report is a saved search (optionally
scheduled via cron). An alert is a scheduled saved search with a trigger
condition (number of results / hosts / sources / custom), a comparator and
threshold, and optional actions. New reports/alerts are created under your profile.
Search
query_splunk runs SPL via search/jobs/export (runs as you; Splunk enforces data
access). Consult splunk_docs (topic spl) before composing queries — SPL is
version-specific.
Session keep-alive
While the server runs, a background loop pings a lightweight REST endpoint to keep
your Splunk session warm and warns early (on stderr) if the SSO cookie goes stale.
Interval: SPLUNK_KEEPALIVE_SECONDS (default 240; 0 disables). Note: SSO cookies
cannot be renewed headlessly — when the session truly expires, run splunk_login
again. The client re-reads the cookie file on every request, so re-logging in
another window is picked up with no restart.
Documentation-first policy
Any agent using this server must read the relevant Splunk docs
before writing SPL or a dashboard definition, then pass docsConsulted: true (and a
docsReference) to the write tools. The tools refuse writes without it. See
.kiro/steering/splunk-platform-authoring.md. This protects accuracy — SPL and the
dashboard schemas are version-specific and must not be guessed.
Setup
npm install
npm run install-browser # one-time: Chromium for Playwright SSOConfigure via your MCP client (mcp.json) env block, or a local .env
(see .env.example). Minimum:
SPLUNK_BASE_URL=https://splunk.<your-org>.comThen, in your client, run the splunk_login tool (or npm run login) and complete
SSO in the window that opens.
Example mcp.json entry
{
"mcpServers": {
"splunk-platform": {
"command": "node",
"args": ["c:/MCP Projects/splunk-platform-mcp/src/index.js"],
"env": {
"SPLUNK_BASE_URL": "https://splunk.example.com"
}
}
}
}Then run the splunk_login tool (or npm run login) and complete SSO. Auth is
SSO-cookie only.
Validate
npm run validate # syntax + config (no network)
npm run validate -- --probe # live: whoami + sample read (after completing login)
npm test # unit tests (format detection, docs gate)Notes
Reading works with any authenticated account. Writing requires your Splunk role to allow it (e.g. the
edit_viewcapability and write access to the target app). If your account is read-only, create/update will be refused by Splunk's ACL — the tools surface that clearly rather than failing with a raw 403.New dashboards are created private (owned by you). Change sharing in Splunk if others need access.
npm
Package name: @svasimahmed283/splunk-platform-mcp. Install or run via MCP:
"args": ["-y", "@svasimahmed283/splunk-platform-mcp@0.1.2"]Repository: Wasim-Shaikh25/splunk-platform-mcp.
Available Tools
17 toolscheck_dashboard_accessA
Check whether you can edit a dashboard BEFORE trying to update it. Reads the Splunk ACL and reports can_write with a plain-language verdict.
Inputs: name (required), app, owner.
create_alertA
Create an alert (scheduled saved search with a trigger condition) under YOUR profile. Requires docsConsulted=true. DOCUMENTATION-FIRST: read the official Splunk docs (use the splunk_docs tool) before writing SPL or a dashboard definition; do not guess syntax or schema.
Inputs: name (required), search (required), docsConsulted (required), app, +10 more.
create_dashboardA
Create a NEW dashboard under YOUR profile (owner = your Splunk user, private by default). Supports Dashboard Studio (JSON) and Classic (Simple XML). Requires docsConsulted=true. DOCUMENTATION-FIRST: read the official Splunk docs (use the splunk_docs tool) before writing SPL or a dashboard definition; do not guess syntax or schema.
Inputs: name (required), definition (required), docsConsulted (required), app, +3 more.
create_reportA
Create a report (saved search) under YOUR profile. Optionally scheduled via cron. Requires docsConsulted=true. DOCUMENTATION-FIRST: read the official Splunk docs (use the splunk_docs tool) before writing SPL or a dashboard definition; do not guess syntax or schema.
Inputs: name (required), search (required), docsConsulted (required), app, +5 more.
get_alertA
Get one alert (saved search with alerting): SPL, schedule, trigger condition, actions, ACL.
Inputs: name (required), app, owner.
get_dashboardA
Fetch one dashboard's full definition (Dashboard Studio JSON or Classic Simple XML), its app/owner/sharing, detected format, and your can_write flag.
Inputs: name (required), app, owner.
get_reportA
Get one report (saved search): its SPL, schedule, ACL, and full config.
Inputs: name (required), app, owner.
list_alertsA
List saved searches classified as alerts (scheduled + triggering) that you can see.
Inputs: app, count, owner, query.
list_dashboardsA
List dashboards you can see (Splunk enforces visibility by your account). Filter by app, owner, or a name/label substring. Wildcards: omit app/owner to search all.
Inputs: app, count, owner, query.
list_reportsA
List saved searches classified as reports (not alerts) that you can see. Filter by app/owner/name.
Inputs: app, count, owner, query.
query_splunkA
Run an SPL search and return results (runs as you; Splunk enforces data access). Use the splunk_docs tool (topic 'spl') to verify command syntax before composing queries. DOCUMENTATION-FIRST: read the official Splunk docs (use the splunk_docs tool) before writing SPL or a dashboard definition; do not guess syntax or schema.
Inputs: query (required), latest, earliest, maxResults.
splunk_docsA
Return the official Splunk documentation references you MUST consult before authoring SPL or dashboards. Call this first when writing queries or dashboard definitions. Optional topic: spl | studio | classic | rest.
Inputs: topic.
splunk_loginA
SSO login in a browser (Playwright); saves cookies for the Splunk REST API. Auth is SSO-cookie only. If IdP redirects or automation block the session, delete the reported cookie file and retry, and complete SSO fully in the opened window.
Inputs: no named parameters.
update_alertB
Update an existing alert. Pre-checks edit access; refuses if can_write is false. Requires docsConsulted=true. DOCUMENTATION-FIRST: read the official Splunk docs (use the splunk_docs tool) before writing SPL or a dashboard definition; do not guess syntax or schema.
Inputs: name (required), search (required), docsConsulted (required), app, +11 more.
update_dashboardA
Update an EXISTING dashboard's definition. Pre-checks your edit access and REFUSES if Splunk ACL denies write. Requires docsConsulted=true. DOCUMENTATION-FIRST: read the official Splunk docs (use the splunk_docs tool) before writing SPL or a dashboard definition; do not guess syntax or schema.
Inputs: name (required), definition (required), docsConsulted (required), app, +3 more.
update_reportA
Update an existing report. Pre-checks edit access; refuses if can_write is false. Requires docsConsulted=true. DOCUMENTATION-FIRST: read the official Splunk docs (use the splunk_docs tool) before writing SPL or a dashboard definition; do not guess syntax or schema.
Inputs: name (required), search (required), docsConsulted (required), app, +6 more.
whoamiA
Show the Splunk username the current session authenticates as. New dashboards are created under this profile.
Inputs: no named parameters.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 17 tools
Most tools map cleanly to distinct resources and actions (dashboards, reports, alerts, auth, docs, query). The only mild overlap is get_dashboard already including a can_write flag while check_dashboard_access exists as a separate pre-check tool.
The set follows a clear verb_noun pattern for the main resources (list_dashboards, get_dashboard, create_report, update_alert). A few outliers like splunk_login, splunk_docs, whoami, and query_splunk break the pattern but are still recognizable.
17 tools is on the heavier side but justifiable given three resources (dashboards, reports, alerts) plus auth, query, and docs. Each tool has a clear role, though the count is at the upper edge of ideal.
Core lifecycle is covered for dashboards, reports, and alerts (list/get/create/update), but there are no delete tools for any resource, and sharing/permission management is only checkable, not settable. Search is present but saved search lifecycle lacks deletion or duplication.
Related MCP Connectors
Query and manage SquaredUp dashboards, data streams, the knowledge graph, and KPIs (US region).
Dashboards as data. Author, validate, render, and share dvt dashboards from any MCP client.
Query and manage SquaredUp dashboards, data streams, the knowledge graph, and KPIs (EU region).
Read Klips, dashboards, data sources and users; refresh data source instances in Klipfolio.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables AI assistants to interact with Splunk Enterprise and Splunk Cloud instances through standardized MCP interface. Supports executing SPL queries, managing indexes and saved searches, listing applications, and retrieving server information with flexible authentication options.-
- FlicenseNot gradedqualityNot gradedmaintenanceEnables interaction with Tableau Cloud and Tableau Server through the REST API, supporting workbook management, view queries, extract refreshes, and content search operations.-
- FlicenseNot gradedqualityNot gradedmaintenanceEnables users to interact with Splunk instances to execute SPL queries, manage saved searches, and monitor system health. It provides a clean interface for listing dashboards, indexes, and logs through the Model Context Protocol.512 npm-
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to search, manage, and analyze data in Splunk instances through the Model Context Protocol. Supports SPL queries, index management, alerts, dashboards, and more.2,353 PyPI1Apache 2.0