.env.example

# Copy to .env (git-ignored) and fill in. kassi loads .env on `kassi serve`.

# Splunk MCP Server (official Splunkbase app 7931). Endpoint is the app's
# streamable-HTTP path on the management port; token is the encrypted MCP token
# generated in the app UI.
KASSI_SPLUNK_MCP_ENDPOINT=https://localhost:8089/services/mcp
KASSI_SPLUNK_TOKEN=

# Set to 1 only for a local Splunk with a self-signed cert (skips TLS verification
# in the mcp-remote bridge). Never set against a real CA-backed endpoint.
KASSI_SPLUNK_INSECURE=1

# k6 MCP server (defaults to the built-in `k6 x mcp` subcommand on k6 2.0+)
# KASSI_K6_CMD=mcp-k6      # use the standalone binary instead
# KASSI_K6_DOCKER=1        # or run via Docker

# Model backend (default: local Ollama running IBM Granite 4.1). Granite natively grounds
# the analysis on the evidence documents kassi passes, which keeps the writeup to the facts.
KASSI_LLM=ollama
KASSI_MODEL=granite4.1:8b
OLLAMA_HOST=http://localhost:11434     # point at the host running Ollama (e.g. a LAN box: http://<lan-host>:11434)
# Use the Claude API instead:
# KASSI_LLM=anthropic
# KASSI_MODEL=claude-haiku-4-5
# ANTHROPIC_API_KEY=

# Output dashboard: kassi publishes each run's verdict + metrics to Splunk via HEC.
# Run scripts/setup_dashboard.py to create the index, token, and dashboard, then paste the token.
KASSI_HEC_TOKEN=
KASSI_HEC_URL=http://localhost:8088
KASSI_RUN_INDEX=kassi_runs