Enter your SAP Commerce Cloud CCv2 credentials below. All values are stored encrypted per workspace.
Team members
Name
Email
Role
Last login
Status
SetupConnect your CCv2 environment
Week 1–2 · CCv2 real API calls
4
connectors
0
connected
0
failed
—
pass rate
1
Connectors
2
Schema
3
Baseline
4
Checklist
Field → Agent mapping
Loading…
Field
Source
Agent
Tier
Status
KPI Baseline Capture
Pulls real KPI values from your connected CCv2 endpoints. Run daily for 14 days to build statistical normal ranges for all agents.
Baseline history0 / 14 days
✅
Setup complete! Cleared to start Live alerts.
Completion checklist0 / 12
Live alertsReal-time alerts from your CCv2
🔒 Observation mode
0
agents active
0
real alerts
—
FP rate
0
observations
—
goal FP <5%
Live CCv2 monitoring (Dynatrace-fed)
Agent
Enable
Threshold
Status
🩺
Commerce health agents need the jarvishealthapi extension
Cache hit rates, stuck orders by time-in-status, catalog quality, Solr indexer lag — these have no external CCv2 API. Install the jarvishealthapi SAP Commerce extension to feed Commerce health agents real data.
# 1. Copy the extension into your SAP Commerce project
cp -r jarvishealthapi/ <hybris>/bin/custom/
# 2. Add to localextensions.xml
<extension name="jarvishealthapi"/>
# 3. Set token in local.properties
jarvis.api.token=<your-secret-token>
# 4. Build via CCv2 pipeline
ant clean all
# 5. Verify (unauthenticated health probe)
curl <node-url>/jarvismeta/metrics/health
Then open Credentials → enter Base URL + Token → click Test.
Commerce health (jarvishealthapi)
Agent
Enable
Threshold
Status
False positive rates — goal <5%—
Predict gate
Alert feedlive
Enable agents to see alerts
Tuning guide
↑ Too many alerts?
Raise the threshold value.
↓ Missing real issues?
Lower threshold — more sensitive.
✓ Target
FP <5% for 3 consecutive days.
ForecastsForecast load & spot anomalies
Week 6–8
0
models trained
—
avg accuracy
0
days ingested
0%
readiness
Historical data ingestion — 30-day feed
// Click Run ingest to pull 30-day historical data
Tier 1 model training
Service response time — 24h forecastTrain models first
Scenario controls
Traffic multiplier1.0×
Promo eventOFF
Pod count8
Prediction
Peak response time—
SLA breach windows—
Recommended pods—
Anomaly patterns — from 30-day history
Run a scan to detect patterns
Automate gate
⚡
Guided incident playbookslive
Guided playbooks for managed SAP Commerce Cloud — every step labelled by who actually pulls the lever.
0
automated actions
0
recommendations
0
approved
—
avg time to fix
0%
readiness
Response playbooks
Activation checklist
Connect your Git provider
Jarvis works with Bitbucket, GitHub, GitLab, and Azure DevOps. Select your provider below and enter a read-only token — Jarvis will scan repos, PR diffs, and ImpEx files without write access.
Bitbucket
API Token / Access Token
GitHub
Personal Access Token
GitLab
Access Token
Azure DevOps
Personal Access Token
⚠️ App Passwords are deprecated (fully disabled June 9, 2026).
Use an API Token: Atlassian account → Security → API tokens → Create API token (select Bitbucket, add Repositories:Read + Pull requests:Read)
Personal token requires your Atlassian account email.
Leave empty if using a Workspace or Repo Access Token
Connected
Loading repos…
Scan entire repository
Review Pull Request
Find & validate all ImpEx files
Scan history
Loading…
Log Intelligence — Dynatrace Telemetry Analysis
Analyses SAP CCv2 health using Dynatrace Problems, Metrics, and Entity topology — no logs.read required. Claude correlates open problems with metric anomalies and service dependencies to produce root cause analysis and fix steps.
ImpEx Validator — validate before import
Runs 10 static rules (missing separators, hardcoded passwords, date formats, etc.) + Claude AI semantic analysis. Returns a structured report with go/no-go recommendation.
Deployment Diff Agent — detect breaking changes
Take a metrics snapshot before deploying, then another after. Compare to automatically detect error rate spikes, response time regressions, new exception types, and throughput drops.
Snapshot workflow
1 Take pre-deploy snapshot → note the ID
2 Deploy your build to CCv2
3 Wait 5 min → take post-deploy snapshot
4 Click Compare → get verdict: STABLE / ROLLBACK RECOMMENDED
Code Review Agent — CCv2 best practices
Paste Java, Groovy, items.xml, or Spring XML code. Runs 12 static pattern rules (Jalo usage, Hibernate bypass, System.exit, empty catch blocks, etc.) + Claude AI review for logic errors, thread safety, and CCv2-specific constraints.
Data Integrity Agent — OCC deep validation
Queries your OCC REST API to detect: zero/negative prices, missing images/descriptions, stuck orders, expired promotions still active, negative stock levels, and catalog inconsistencies.
Reads application logs from the SAP-provisioned Azure Blob Storage container, clusters recurring ERROR/WARN patterns, and uses Claude to identify root causes and fix steps.
⏱
Logs are not real-time.
SAP writes CCv2 application logs to Azure Blob Storage in batches — entries typically appear 5 to 30 minutes after they occur, depending on your environment's log flush interval.
This makes SupportAgent suitable for pattern discovery and post-incident analysis, not live alerting.
For real-time incident detection use the Dynatrace connector (Monitor tab) — it polls every 60 seconds.
Jarvis automatically scans logs on your plan schedule and alerts when new critical issues are detected.
Slack
PagerDuty routing key
How far back to scan blob files. Note: logs are not real-time — SAP flushes CCv2 logs to Azure Blob every 5–30 minutes. Use 1–2 hours for recent incidents, 12–24 hours for full pattern discovery.
How many of the most frequent error patterns to include in the Claude analysis. 10 is a good default — covers the critical issues without exceeding Claude's context. Increase to 20–30 when you want a full health audit. Patterns are ranked by frequency (most recurring first).
⬇ Ingest logs — downloads recent log files from the Azure Blob container, parses entries, and clusters similar messages by fingerprint. Run this first. 🤖 Analyse with Claude — sends stored patterns to Claude AI for root-cause analysis, business impact scoring, step-by-step fix recommendations, and identifies which fixes can be auto-generated. Requires a previous ingest run.
