Quickstart Guide

This guide gets you up and running with KLA Digital. You will install a native telemetry SDK, register a governed agent, run an execution with human checkpointing, and export a cryptographically sealed evidence bundle.

---

Run the KLA CLI

The KLA CLI currently runs from this monorepo. From the repository root, build it with Node.js 20.x:

pnpm install --frozen-lockfile
pnpm kla --help
alias kla='pnpm --silent kla'

The repository wrapper rebuilds the CLI when a source file is newer than its generated output.

---

Step 1: Install a Telemetry SDK

KLA integrates with standard frameworks using native OpenTelemetry instrumentation. Choose your language:

Node.js

pnpm add @kla-digital/otel-node

Python

pip install kla-otel-python

---

Step 2: Point OTLP to KLA

Configure your service environment variables to route telemetry spans through the KLA Collector. You can fetch your specific API endpoint and tenant tokens from the KLA Console.

OTEL_SERVICE_NAME=claims-triage-agent
OTEL_EXPORTER_OTLP_ENDPOINT=https://api.kla.digital/collector
KLA_TENANT_ID=kpmg-reporting
KLA_API_KEY=kla_sec_prod_8f2d89a2
KLA_PII_MASK=true

---

Step 3: Register your Agent

Register a new agent declaration by sending a manifest to the KLA Control Plane. This defines the agent's name, description, parameters, and allowed tool lists:

curl -X POST https://api.kla.digital/execution/v1/agents \
  -H "Authorization: Bearer $KLA_API_KEY" \
  -H "x-kla-tenant-id: $KLA_TENANT_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "claims-bot",
    "description": "Triages inbound customer claims with human-approval gates",
    "model": "gpt-4o-mini",
    "temperature": 0.2,
    "tools": ["search_claims_db", "process_refund"]
  }'

The server response will output a stable agentId (e.g. agt_9f81a7) and a compiled manifestHash representing your configuration state.

---

Step 4: Trigger a Governed Execution

Start a new workflow run by routing your execution request through the KLA Engine. KLA evaluates active policies in real-time. If the input prompt triggers a safety rule (e.g. initiating a refund over $1000), KLA pauses execution and places the task on the Decision Desk for review.

curl -X POST https://api.kla.digital/execution/v1/executions \
  -H "Authorization: Bearer $KLA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agentId": "agt_9f81a7",
    "input": { "prompt": "Process refund for claim #9921 for $1250.00" },
    "budgets": { "maxSteps": 50, "maxTokens": 20000, "maxCostUsd": 5 },
    "flags": { "euOnly": true }
  }'

You can listen for execution updates and human-in-the-loop decisions by subscribing to the WebSocket stream at: wss://api.kla.digital/execution/ws

---

Step 5: Export a Sealed Evidence Bundle

After executions conclude, you can download evidence sheets utilizing the kla command-line utility. These packages are cryptographically signed and contain the bundle's Merkle root, ledger anchors, and audit entries matching the exact traces of the workflow.

# Export the last 30 days to a local directory
kla evidence export --tenant $KLA_TENANT_ID --days 30 --out ./evidence

# Export an exact time window
kla evidence export --tenant $KLA_TENANT_ID \
  --start-date 2026-07-01 \
  --end-date 2026-07-31 \
  --out ./evidence

The resulting Sealed Evidence Bundle is ready for delivery to auditors or compliance officers.