Skip to main content
Installation

Requirements

Splunk versions, the two packages, LLM endpoints, outbound network, browsers.

This page lists the environment conditions to confirm before installation.

Splunk

ItemRequirementNotes
VersionSplunk Enterprise 10.0 or laterThe requests engine runs on every version
Agentic engine10.2 or later, Python 3.13The official splunklib.ai Agent exists only on this combination. Anything older falls back to the requests engine and the AI settings page states why (python_too_old, splunklib_ai_missing)
KV storeEnabledConversations, knowledge base, triage results, asset tables, reports and preferences all live there
TopologySingle search head or search-head clusterInstalls on the search head; nothing on indexers
Enterprise SecurityOptionalWith ES, live alerts read index=notable and verdicts write back to the notable. Without ES the alert source can be any index
Splunk CloudNot supported yetThe app's online update downloads and extracts a .spl and restarts Splunk, which would not pass Splunk Cloud vetting. There is no Cloud build

The two packages

PackageContentsUse
RST-AI-Copilot-for-Splunk-<version>-fallback.splPure Python, requests engineSplunk Enterprise where the selfcontained package cannot be used: Windows or ARM (aarch64) hosts, or environments that do not allow compiled components
RST-AI-Copilot-for-Splunk-<version>-selfcontained.splsplunklib.ai and dependencies bundled (about 120 MB of cp313 wheels), licence trust anchor compiled to .soSplunk Enterprise on Linux x86_64 only; the agentic engine needs 10.2 or later
RST-AI-Copilot-for-Splunk-<version>-splunkbase.splSame as the fallback package, with the in-app update offFor distribution through Splunkbase; updated through Apps > Manage Apps

Both packages have the same features. The difference is the agentic engine: it lets the model call read-only tools to probe the real schema, dry-run its SPL and gather evidence around an alert. The requests engine answers the same prompts in one turn without probing.

A 10.2 or later host running the fallback package can later run scripts/vendor_agentic.sh on the search head to install the stack into bin/lib/, which is equivalent to the selfcontained package.

The splunkbase package differs from the fallback package in two ways only: default/copilot.conf sets [update] enabled = 0, so the in-app update is off, and app.conf sets check_for_updates = 1, so new versions show in Splunk's Manage Apps; the update card in the app is replaced by a notice. See Online update.

LLM

Any OpenAI-compatible chat-completions endpoint. Verified: Volcengine Ark coding plan (ark-code-latest, doubao-seed-*, kimi-k2.7-code, deepseek-v4-*, glm-5.3), DeepSeek, Alibaba DashScope, OpenAI, OpenRouter, local Ollama and vLLM.

RequirementNotes
Context of at least 32kInvestigation and triage prompts carry evidence; a short context truncates them
Function calling not requiredThe agentic engine uses splunklib.ai's own protocol
Embeddings endpoint optionalSemantic retrieval is off by default and the knowledge base falls back to keyword retrieval

Network

DirectionTargetPurpose
OutboundLLM endpoint 443Every AI call
Outboundlicense.reallysec.com 443Online activation, heartbeat, revocation list. Not needed with an offline licence
Outbound (optional)Threat-intel vendor, ServiceNow and Jira, Feishu, DingTalk, WeCom, Slack, Teams, SMTP, syslog, webhookEach egresses only once enabled in copilot.conf or in the UI; all off or unconfigured by default
InboundNoneEverything goes through Splunk Web on 8000

The MCP server (1.9.0 and later) listens on 127.0.0.1:18100 by default. Other machines need a TLS reverse proxy in front.

Browser

The last two major versions of Chrome, Edge, Firefox and Safari. The UI is laid out for 1280 px and wider; below 1024 px the sidebar collapses.

On this page