Panel 1 — Automation Architecture · Updated June 2026
Zero-Prompt Autopilot Workflows
AI acts on triggers without anyone asking. A Teams message becomes a Jira ticket. A failing build gets diagnosed. A PR gets reviewed. All automatically. Built on MCP 2025-06-18 (Streamable HTTP, OAuth 2.0) + Spring AI 1.1.
// 2026 model selection for automation
Classification / Routing / Ticket enrichment / Build diagnosis → Haiku 4.5 ($1/$5 per 1M — 3x cheaper than Sonnet)PR review / Complex analysis / Architecture checks → Sonnet 4.6 ($3/$15 per 1M)
Sprint-end docs / Nightly batch jobs → Batch API (50% off any model)
Automation system prompts (repeated every event) → Prompt cache — 1h TTL (~90% cost reduction on cached tokens)
// EVENT SOURCES → AI BRAIN → ACTIONS
Microsoft Teams ──message──► MCP Server ──► Claude analyses ──► Jira ticket created + replied in thread
Jira new ticket ──webhook──► MCP Server ──► Claude enriches ──► ACs added, gaps flagged, estimate suggested
GitHub PR opened ──event──► MCP Server ──► Claude reviews ──► Review comment posted, issues labelled
CI build fails ──webhook──► MCP Server ──► Claude diagnoses ──► Fix suggestion on PR + Slack alert
Security scan ──report──► MCP Server ──► Claude triages ──► False positives removed, CRITICAL flagged
Perf alert fires ──Grafana──► MCP Server ──► Claude investigates ──► Root cause + code suspect in Slack
Test results ──JUnit XML──► MCP Server ──► Claude analyses ──► Failure patterns, flaky tests, coverage gaps
Standup time ──cron 9am──► MCP Server ──► Claude reads Jira ──► Pre-fills standup per team member in Teams
01
TRIGGER
Teams msg, Jira webhook, GitHub event, Grafana alert, cron schedule, CI result
02
MCP SERVER
Your Spring Boot MCP server receives the event. Routes to the right AI handler.
03
CLAUDE ACTS
Claude reads context from Jira, Confluence, GitHub, code. Reasons and decides action.
04
ACTION
Creates ticket, posts comment, updates doc, fires alert, adds review — all via APIs.
05
DASHBOARD
All AI actions logged in real-time. Full audit trail. Human can override at any step.
⚡ Token efficiency for automated tasks: Use Haiku 4.5 ($1/$5/1M) for classification, routing, ticket enrichment, build diagnosis. Use Sonnet 4.6 ($3/$15/1M) for PR reviews, complex analysis. Use Batch API (50% off) for sprint-end doc generation, nightly analysis. Enable prompt caching — automation system prompts are large and repeated constantly. A team running 100 automations/day can save 70%+ vs naive implementation.
Fully Automated (no human needed)
Semi-Auto (AI drafts, human approves)
AI-Assisted (human initiates, AI accelerates)
Teams Message → Jira Ticket
Someone asks a question → AI creates the ticket
TRIGGER
Teams message in #dev-help or #questions contains a question or a problem description. The bot monitors the channel 24/7.
AI DOES
Reads the message. Classifies: is it a bug report, feature request, question, or blocker? Searches Confluence and Jira for existing answers. If novel → creates a Jira ticket with title, description, and suggested priority. Replies in Teams thread with the Jira link + summary of any existing docs that might already answer it.
OUTPUT
Jira ticket created · Teams reply with answer or ticket link · Zero manual effort for the team member who asked
Teams Bot System PromptAUTO
You monitor the Teams channel. When a message arrives:
1. Classify: bug / feature / question / blocker / announcement
2. Search Confluence for: [search_confluence(query)]
3. Search Jira for existing tickets: [search_jira(query)]
4. If answered in docs: reply with the Confluence link + 2-sentence summary
5. If no answer exists: create_jira_ticket(title, description, priority, labels)
6. Reply in thread: "I've created [TICKET-NNN] for this. [summary]. [link]"
Never create duplicate tickets. Always check first.
New Jira Ticket → AI Enrichment
Ticket created → AI fills in missing fields
TRIGGER
Jira webhook: issue_created for any ticket in your project board
AI DOES
Reads the ticket. Fetches related Confluence pages. Adds: missing acceptance criteria, suggested story point estimate (1/2/3/5/8), recommended labels/components, links to related tickets, and a "Gaps identified" comment flagging any ambiguities before the team picks it up.
OUTPUT
Ticket updated with AC, estimate, labels · Comment added listing gaps · BA/PO notified if gaps are critical
Jira Enrichment PromptAUTO
New Jira ticket received: {ticket_json}
Team stack: Java/Spring Boot/React/PostgreSQL/AWS
1. Analyse for missing acceptance criteria → add_acceptance_criteria()
2. Estimate story points based on description + similar past tickets → update_story_points()
3. Find gaps: what's ambiguous, missing, or unclear?
4. Post comment: "AI Pre-Review:\n✓ ACs added\n⚠ Gaps: [list]\n📊 Estimate: [N] points"
5. If CRITICAL gaps: tag @BA @PO in comment
PR Opened → Auto Code Review
Every PR gets an AI review before human eyes
TRIGGER
GitHub/GitLab webhook: pull_request opened or synchronize
AI DOES
Reads all changed files. Runs production readiness check: security, performance, test coverage, naming, error handling. Posts inline comments on specific lines for issues. Adds summary comment: APPROVE / REQUEST_CHANGES with severity breakdown.
OUTPUT
Inline review comments on changed lines · Summary comment with MUST FIX / SHOULD FIX · Label added: needs-human-review (after AI approves)
GitHub Action YAML snippetCI/CD
on: [pull_request]
jobs:
ai-review:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- name: AI Code Review
run: |
curl -X POST $MCP_SERVER/review \
-d '{"pr_number":"${{github.event.number}}",
"repo":"${{github.repository}}",
"changed_files":"${{steps.files.outputs.all}}"}'
CI Build Fails → Auto Diagnosis
Build breaks → AI finds why before developer looks
TRIGGER
CI webhook: build_failed — GitHub Actions, GitLab CI, Jenkins
AI DOES
Reads the full build log. Identifies the root cause (compile error, test failure, dependency issue, OOM). Finds the relevant code from the repo. Posts diagnosis as a PR comment with: root cause in plain English, the specific failing code highlighted, and a suggested fix.
OUTPUT
PR comment: "Build failed: [root cause in 1 sentence]. Likely in [ClassName.java line N]. Suggested fix: [code snippet]" · Teams alert if on main/release branch
Build Failure Diagnosis PromptAUTO
CI build failed. Analyse and diagnose.
Build log (last 200 lines):
{build_log_tail}
Changed files in this PR: {changed_files_list}
1. What is the ROOT CAUSE? (1 sentence)
2. Which file and line number is responsible?
3. Is this: compile error / test failure / dependency / config / OOM?
4. What is the EXACT FIX? (provide code)
5. Is this a flaky test or a real failure?
Post to PR as a comment. Tag the PR author.
Security Scan → AI Triage
SAST/DAST results prioritised, false positives removed
TRIGGER
SAST tool result (SonarQube, Snyk, Checkmarx, Trivy) or DAST result (OWASP ZAP)
AI DOES
Reads every vulnerability. Removes false positives (contextual analysis — e.g. SQL injection false positive in a parameterised query). Ranks remaining by actual risk (CVSS + context). Creates Jira tickets for CRITICAL/HIGH. Adds to security backlog for MEDIUM. Ignores LOW with a comment explaining why.
OUTPUT
Jira tickets for real CRITICALs · Security dashboard updated · False positives marked with explanation · Security team notified of new CRITICALs
9am Cron → Standup Pre-fill
AI reads Jira activity, writes everyone's standup
TRIGGER
Cron job at 8:45am every working day
AI DOES
Reads Jira activity log for each team member (last 24h). Reads PR comments, commits, review activity. Drafts a standup note per person: "Yesterday: [activity]. Today: [in-progress tickets]. Blockers: [any tickets stuck > 2 days without update]." Posts to Teams standup channel as a threaded message, each person gets their own card to edit and reply to.
OUTPUT
Teams standup channel: pre-filled card per team member · Takes 30 seconds to confirm instead of 5 minutes to prepare · Blockers surfaced automatically
Grafana Alert → Performance Investigation
Latency spike → AI finds the code and query suspect
TRIGGER
Grafana webhook: alert firing for p95 latency, error rate, or DB query time threshold
AI DOES
Reads the alert: which service, which endpoint, which metric. Fetches recent slow query logs. Reads the relevant service code from GitHub. Correlates with recent deployments (was there a deploy in the last 2h?). Posts: suspected cause, whether it's code regression or data volume growth, and the specific method/query to investigate.
OUTPUT
Teams alert: "[Service] p95 spiked to 4.2s. Suspected: N+1 query in OrderService.findByUser() added in PR #342 (deployed 1h ago). Confirm with: EXPLAIN ANALYZE [query]"
Sprint End → Auto Documentation
Sprint completes → all docs updated automatically
TRIGGER
Jira webhook: sprint_closed
AI DOES
Reads all completed tickets. Reads merged PRs. Updates API docs in Confluence for any changed endpoints. Generates sprint summary (completed, not completed, velocity). Drafts release notes for stakeholders. Updates architectural diagrams if new services or integrations were added.
OUTPUT
Confluence updated: API docs, architecture page · Sprint summary posted to Teams #releases · Draft release notes ready for PM to approve
Confluence Page Edited → Code Sync Check
Doc changed → AI checks if code matches
TRIGGER
Confluence webhook: page_updated for API or architecture pages
AI DOES
Reads the updated Confluence page. Reads the corresponding code from GitHub. Checks for discrepancies: API paths that exist in docs but not code (or vice versa), data types that don't match, behaviours described but not implemented. Posts a comment on the Confluence page flagging any gaps.
OUTPUT
Confluence comment: "⚠ Doc-Code Drift Detected: DELETE /api/orders/{id} exists in docs but not in OrderController.java · Notify: @techwriter @leaddev"
Implementation
Building the Automation Hub: MCP Server
One Spring Boot MCP server (Spring AI 1.1 + MCP 2025-06-18 spec) receives all webhooks and routes to Claude. Streamable HTTP transport. OAuth 2.0 auth. Tool annotations (read-only/destructive) on every tool.
AutomationHub.java — Spring Boot MCP Server skeletonJava / Spring AI 1.1 MCP (Nov 2025)
@SpringBootApplication
@McpServer
public class AutomationHub {
// ── TOOLS: what Claude can do (MCP 2025-06-18 — use @ReadOnlyTool / @DestructiveTool annotations) ──
@Tool(value="Creates a Jira ticket. Use when a Teams message describes a new bug, feature, or task that doesn't have an existing ticket.", annotations={@ToolAnnotation(type="destructive")})
public String createJiraTicket(String title, String description, String priority) { ... }
@Tool("Searches Jira for existing tickets matching the query. Always call this BEFORE creating a new ticket.")
public List searchJira(String query) { ... }
@Tool("Searches Confluence pages for documentation matching the query.")
public List searchConfluence(String query) { ... }
@Tool("Posts a reply to a Teams message thread.")
public void replyTeams(String threadId, String message) { ... }
@Tool("Posts a review comment on a GitHub PR at a specific file and line.")
public void postPrComment(String repo, int pr, String file, int line, String body) { ... }
@Tool("Updates a Jira ticket: sets story points, acceptance criteria, labels, or status.")
public void updateJiraTicket(String ticketId, Map fields) { ... }
@Tool("Posts to a Teams channel. Use for alerts and notifications.")
public void postToTeams(String channel, String message) { ... }
// ── RESOURCES: read-only context (no side effects, MCP 2025-06-18) ─────────
@McpResource("jira://tickets/{project}")
public List getProjectTickets(@PathVariable String project) { ... }
@McpResource("github://diff/{repo}/{pr}")
public String getPrDiff(@PathVariable String repo, @PathVariable int pr) { ... }
@McpResource("metrics://latency/{service}")
public LatencyMetrics getServiceMetrics(@PathVariable String service) { ... }
// ── WEBHOOKS: what triggers Claude ────────────────────
@PostMapping("/webhook/teams")
public void onTeamsMessage(@RequestBody TeamsEvent event) {
claude.process(TEAMS_SYSTEM_PROMPT, event.getMessage());
}
@PostMapping("/webhook/jira")
public void onJiraEvent(@RequestBody JiraEvent event) {
if ("issue_created".equals(event.getType()))
claude.process(JIRA_ENRICHMENT_PROMPT, event.getIssue());
}
@PostMapping("/webhook/github")
public void onGitHubEvent(@RequestBody GitHubEvent event) {
if ("pull_request".equals(event.getType()))
claude.process(PR_REVIEW_PROMPT, event.getPullRequest());
}
}
Panel 2 — All Team Roles · 11 Roles · Updated June 2026
AI Playbook for Every Role
Every person on a large enterprise team — from Business Analyst to Security Engineer — gets specific AI workflows tailored to their daily work.
Fully Automated
Semi-Automated
AI-Assisted Prompt
📊
Product Owner (PO)
Backlog, stories, priorities, release planning
User story generation: meeting recording/notes → AI extracts user stories with ACs automatically
AUTOBacklog grooming: every ticket analysed for completeness before sprint planning — gaps flagged overnight
AUTOPriority scoring: "Score this backlog by business value, urgency, and effort. Suggest the top 10 for next sprint."
PROMPTRelease notes: "Generate stakeholder-friendly release notes from these merged tickets: [list]"
PROMPTStory splitting: tickets estimated >8 points are automatically flagged with split suggestions
AUTOImpact analysis: "If we delay feature X by 2 sprints, what dependent tickets are affected? What's the customer impact?"
PROMPT🔍
Business Analyst (BA)
Requirements, gap analysis, process flows
Requirements extraction: paste stakeholder email/meeting notes → AI extracts structured functional requirements
PROMPTGap analysis: "Compare these business requirements against the existing system docs. What's not covered?"
PROMPTBusiness rules extraction: AI reads legacy code and Confluence → extracts implicit business rules into a structured register
SMARTProcess flow generation: "Generate a BPMN-style process flow for this user story as a text diagram"
PROMPTAs-is / to-be analysis: "Here's the current process. Here's the proposed change. List all business rule impacts."
PROMPTConfluence sync: when code changes, AI automatically flags BA docs that reference the changed behaviour
AUTO🏗️
Solution Designer (SD)
Architecture, design, technical leadership
ADR generation: describe a tech decision → AI writes formal ADR with tradeoffs and alternatives
PROMPTArchitecture critique: "You are a principal engineer. Find flaws in this design before we build it: [design description]"
PROMPTC4 model generation: AI reads all repos and generates context/container/component diagrams as text — auto-updated on new PRs
SMARTSequence diagram: "Generate a PlantUML sequence diagram for the order placement flow across @folder [order-service] @folder [payment-service]"
PROMPTRFC drafting: "Draft an RFC for introducing [technology/pattern] to our stack. Include: rationale, migration path, risks, rollback."
PROMPTTech debt analysis: AI scans all repos monthly → produces tech debt register with effort estimates and risk ratings
SMART💻
Developer (All Levels)
Code, review, debug, test
Ticket → Tests → Code: full TDD pipeline automated — failing tests committed, implementation follows
AUTOPR auto-review: every PR reviewed by AI before human reviewers — MUST FIX issues flagged inline
AUTOLegacy code explanation: "Explain what this 500-line class does and whether it can be safely refactored: @file"
PROMPTBuild failure diagnosis: CI fails → AI posts diagnosis on the PR within 60 seconds
AUTOCross-repo impact: "If I change this method signature, what breaks across @folder [all repos]?"
PROMPTStandup pre-fill: AI reads your Jira activity and drafts your standup every morning at 8:45am
AUTO🧪
Quality Engineer (QE)
Test planning, automation, defect analysis
Test plan generation: new epic created → AI generates full test plan covering functional, regression, and edge cases
AUTOTest case generation: user story accepted → AI generates test cases from ACs, ready to import to Xray/Zephyr
AUTODefect triage: new defect logged → AI reads the code, logs, and stack trace → adds root cause hypothesis and test coverage gap
AUTORegression selection: "Given these changed files, which existing test cases should be included in the regression run?"
PROMPTFlaky test detection: AI monitors CI results over 2 weeks → identifies tests that fail intermittently, files Jira tickets
AUTOAPI test generation: "Generate Postman/REST Assured tests for every endpoint in @file [Controller.java] covering all status codes"
PROMPT🚀
DevOps / Platform Engineer
Infra, CI/CD, deployment, reliability
Pipeline failure diagnosis: CI/CD fails → AI reads logs, identifies stage, posts fix to the team channel
AUTOInfrastructure drift: Terraform plan on schedule → AI compares with IaC → flags manual changes in prod vs config
AUTODeployment risk assessment: "Analyse this deployment plan for risks: services affected, rollback strategy, data migrations, blast radius"
PROMPTTerraform generation: "Write Terraform for an ECS service + RDS PostgreSQL + ALB for this Spring Boot app. Follow our existing module patterns @folder [terraform/modules]"
PROMPTCost anomaly: AWS cost spike detected → AI analyses usage vs last week, identifies which service/resource changed
AUTORunbook generation: "Read @folder [k8s/] and generate a complete operational runbook for the on-call team"
PROMPT🛡️
Security Engineer
Vulnerability triage, SAST, threat modelling
SAST triage: SonarQube/Snyk report → AI removes false positives, ranks real vulnerabilities, creates tickets for CRITICAL/HIGH
AUTOSTRIDE threat model: "Perform a STRIDE analysis on this component: [describe]. Include likelihood, impact, mitigation for each threat."
PROMPTSecurity requirement extraction: new epic → AI reads AC and flags implicit security requirements (auth, audit, encryption) that weren't stated
AUTODependency audit: "Analyse @file [pom.xml]. List all dependencies with known CVEs, their severity, and whether an upgrade path exists."
PROMPTPII data flow audit: AI reads all repos → maps where PII data flows, is stored, and whether it's encrypted — weekly report
SMARTPen test prep: "Generate an attack surface summary for this API based on @folder [controller/]. What should a pen tester focus on?"
PROMPT⚡
Performance Engineer
Load testing, profiling, bottleneck analysis
Latency alert investigation: Grafana alert fires → AI reads metrics, recent deploys, slow query logs → posts root cause to Teams
AUTOLoad test script generation: "Generate a k6 load test for the order placement flow. Use patterns from @folder [tests/load]. Ramp to 500 concurrent users."
PROMPTJMeter/Gatling scenarios: "Read @file [Controller.java] and generate a Gatling simulation covering all endpoints with realistic request distribution"
PROMPTPerformance regression detection: every PR → AI compares benchmark results against baseline → flags regressions >10% before merge
AUTOProfiling analysis: "Here is a JVM heap dump / thread dump. Find the bottleneck, memory leak, or deadlock: [paste output]"
PROMPTCapacity planning: "Given this growth trend and current infrastructure, when will we hit capacity? What needs scaling first?"
PROMPT📈
Engineering Manager
Delivery, team health, risk management
Sprint health monitor: daily cron reads Jira → flags tickets at risk (stuck > 3 days, estimate burned through, blockers unresolved)
AUTOWeekly status report: every Friday AI generates stakeholder status report from Jira velocity, PRs merged, incidents
AUTODelivery risk assessment: "Given our velocity trend and remaining backlog, will we hit the [date] milestone? What's at risk?"
PROMPTRetro analysis: "Analyse this sprint data: completed/not-completed/incidents. Generate retro talking points and action items."
PROMPTTeam health metrics: AI monitors PR review times, build stability, incident frequency → weekly team health score in Teams
SMARTDependency mapping: "Map the cross-team dependencies for this release. Which teams are blockers? What's the critical path?"
PROMPT🔗
Integration Engineer
APIs, contracts, middleware, data pipelines
API contract generation: "Read @folder [service/src] and generate an OpenAPI 3.0 spec for all endpoints with request/response schemas"
PROMPTContract drift detection: consumer publishes new contract → AI compares against provider → flags breaking changes automatically
AUTOIntegration test generation: "Generate WireMock stubs and REST Assured tests for all external API calls in @folder [src/]"
PROMPTData mapping analysis: "Map the data transformation between this source schema and target schema. Flag any field losses or type mismatches."
PROMPTBreaking change detection: API schema change in PR → AI checks all known consumers → flags which will break
AUTO📚
Technical Writer / Documentation
Docs, runbooks, release notes, API guides
API doc generation: every merged PR with API changes → AI auto-updates API reference in Confluence
AUTORelease notes: sprint closed → AI drafts release notes for technical and non-technical audiences
AUTORunbook generation: "Read @folder [service/] and write an operations runbook: startup, config, common errors, escalation"
PROMPTDoc-code drift detection: code changes merged → AI checks if Confluence pages referencing that code are still accurate
AUTOOnboarding docs: "Generate a developer onboarding guide for @folder [service/]. Include: local setup, key concepts, common gotchas."
PROMPTPanel 3 — Testing Phases · Updated June 2026
Making ST and SIT a Breeze
System Testing and System Integration Testing are where most delivery delays happen. AI reduces prep time from days to hours, finds defect patterns before humans do, and keeps environments ready automatically.
🟣 System Testing (ST)
AI Test Plan from Epic
Epic created → AI generates full ST test plan in Confluence: scope, objectives, entry/exit criteria, test types, schedule
AUTOTest Case Generation from ACs
Every AC in every user story → AI creates test case with steps, data, expected results. Ready for Xray/Zephyr import.
AUTORegression Selection
Changed files listed → AI selects which existing test cases must run. Filters out irrelevant tests. Cuts regression suite by 60%.
SMARTDefect Triage & Clustering
Defects logged → AI groups by root cause, identifies systemic issues, suggests which ones are duplicates, flags critical path blockers
AUTOST Progress Report
Daily at 5pm: AI reads test execution results → generates progress report for PM: % passed, defect density, risk to exit criteria
AUTOExit Criteria Evaluation
AI monitors ST metrics continuously. When all exit criteria are met, posts "ST READY TO CLOSE" to Teams with evidence summary
AUTOFlaky Test Detection
AI monitors pass/fail history per test over the ST cycle → flags tests that flip without code changes as flaky, separates from real failures
SMART🔵 System Integration Testing (SIT)
Integration Scenario Generation
Reads all service APIs and event contracts → generates end-to-end SIT scenarios covering cross-service happy paths and failure modes
SMARTAPI Contract Testing
Reads OpenAPI specs of all services → generates Pact consumer-driven contract tests → runs in CI before SIT begins
AUTOEnvironment Readiness Check
Before SIT begins: AI checks all services are deployed and healthy, all test data seeded, all external stubs configured. Posts readiness report.
AUTOCross-Service Defect Tracing
SIT defect logged → AI reads logs from ALL services involved in the failing flow → identifies which service is the root cause
AUTOTest Data Generation
Reads test scenarios and DB schema → generates realistic test data sets covering happy paths, edge cases, and boundary conditions
SMARTEvent Flow Verification
Reads event topology → generates tests that verify events published by service A are correctly consumed by services B, C, D
SMARTSIT Go/No-Go Assessment
AI monitors all SIT metrics → when all integration tests pass and all CRITICALs are resolved → auto-generates Go/No-Go report for release manager
AUTOPrompt Templates
ST / SIT Prompt Library
Generate ST Test Plan from EpicClaude Chat
Generate a complete System Test plan for this epic.
Epic: [PASTE EPIC TITLE AND DESCRIPTION]
Acceptance criteria: [PASTE ALL ACs FROM CHILD STORIES]
Tech stack: Java/Spring Boot/React/PostgreSQL/AWS
Services involved: [list service names]
Write a formal ST plan with:
1. TEST SCOPE: what IS and IS NOT in scope for this ST cycle
2. ENTRY CRITERIA: what must be true before ST starts (code complete, envs ready, data seeded)
3. EXIT CRITERIA: what must be true to close ST (pass rate %, max open defects by severity)
4. TEST TYPES: functional, regression, boundary, negative, performance (which apply and why)
5. TEST CASES: for each AC, 1-3 test cases with: ID, title, preconditions, steps, expected result
6. TEST DATA: what reference data is needed and how to create it
7. RISKS: what could block this ST cycle (env instability, data issues, dependent services)
8. ESTIMATED DURATION: based on test case count and complexity
Format for import to Confluence. Test cases in table format for Xray/Zephyr import.
SIT Scenario GenerationCursor Chat — with all service repos
Generate comprehensive SIT scenarios for this release.
Services in scope: @folder [order-service/src] @folder [payment-service/src] @folder [notification-service/src]
Event contracts: @folder [event-schemas/]
API specs: @folder [api-specs/]
For each cross-service flow:
1. HAPPY PATH scenario: all services respond correctly
2. PARTIAL FAILURE scenario: what if service B is down when service A calls it?
3. RETRY scenario: transient failure → retry → eventual success
4. COMPENSATION scenario: order placed → payment fails → order must be cancelled
5. EVENT ORDERING scenario: events arrive out of order — is the system consistent?
For each scenario provide:
- Scenario ID and title
- Services involved (sender → receiver)
- Test steps: setup → trigger → expected behaviour per service
- Verification: how to confirm the scenario passed (API response, DB state, event published)
- Defect risk: HIGH / MEDIUM / LOW
SIT Defect Root Cause (Multi-Service)Claude Chat
SIT defect: [DEFECT-NNN] [title]
Scenario that failed: [describe what was tested]
Expected: [what should have happened]
Actual: [what happened instead]
Logs from all involved services:
--- order-service ---
[PASTE LOGS]
--- payment-service ---
[PASTE LOGS]
--- notification-service ---
[PASTE LOGS]
Relevant code: @folder [order-service/src] @folder [payment-service/src]
1. Which service is the ROOT CAUSE? (not just where the error surfaces — where it originates)
2. What is the exact line / method responsible?
3. Is this: code bug / config issue / race condition / missing error handling / data issue?
4. What is the minimum fix?
5. Which existing test should have caught this? If none, what test should be added?
6. Does this affect any other integration flows not yet tested?
Environment Readiness CheckAutomated — runs before SIT
Pre-SIT environment readiness check. Verify all of the following:
Services to check: {service_list}
Expected versions: {version_manifest}
Required test data: {test_data_manifest}
Check and report:
1. DEPLOYMENT STATUS: is each service deployed to SIT env at the correct version?
→ call health_check(service, env) for each
2. DATABASE STATE: are all required reference data and seed data present?
→ call check_test_data(env, manifest)
3. EXTERNAL STUBS: are all third-party API stubs configured and responding?
→ call verify_stubs(env)
4. CONNECTIVITY: can each service reach its dependencies?
→ call connectivity_check(service_graph)
5. CONFIG: are all environment variables set and correct?
→ call config_audit(env, services)
Output: READY / NOT READY
If NOT READY: list exactly what's blocking, which team owns each blocker, and estimated time to resolve.
Post report to Teams #sit-channel.
ST Exit Criteria EvaluationAutomated — daily during ST
Evaluate ST exit criteria. Fetch current metrics:
- Total test cases: {total_tc}
- Passed: {passed_tc} ({pass_pct}%)
- Failed: {failed_tc}
- Blocked: {blocked_tc}
- Not run: {not_run_tc}
- Open defects by severity: CRITICAL:{c} HIGH:{h} MEDIUM:{m} LOW:{l}
Exit criteria defined:
- Pass rate ≥ 95%
- Zero open CRITICAL defects
- Open HIGH defects ≤ 2 (with documented workaround)
- All must-test scenarios executed
Evaluate: PASS / FAIL / CONDITIONAL for each criterion.
Overall ST status: READY TO CLOSE / NOT READY
If CONDITIONAL: what specific conditions must be met?
If NOT READY: what is the critical path to ready? Estimated days?
Post to Teams #st-status and update Jira ST milestone.
// ST / SIT Automation Checkpoint
AI generates test plans automatically when an epic enters "In ST" status
Test cases are auto-generated from ACs and imported to Xray/Zephyr without manual QE effort
Environment readiness check runs automatically before every SIT cycle begins
Defect root cause analysis identifies the responsible service within 5 minutes of logging
Exit criteria evaluation runs daily and posts to Teams — no human needs to collate metrics
Flaky tests are automatically separated from real failures in CI and filed as tech debt tickets
Panel 4 — Real-Time View
Live Automation Dashboard
Every AI action across your team visible in real time. All updates below are live-simulated — connect your MCP server to push real events.
0
AI Actions Today
▲ +12 vs yesterday
0
Tickets Auto-Created
▲ 8 from Teams msgs
0
PRs Auto-Reviewed
● 3 awaiting human
0
Defects Auto-Triaged
▲ 3 CRITICAL flagged
// LIVE ACTIVITY FEED
// ST / SIT PROGRESS
ST — Sprint 42
0%
SIT — Release 2.4
0%
Regression Suite
0%
Contracts (Pact)
0%
// AI BOTS STATUS
teams-botwatching #dev-help
jira-enricheridle
pr-revieweridle
ci-diagnosticsanalysing PR #447
security-triageidle
perf-monitorwatching metrics
st-orchestratorST 87% complete
sit-orchestratorrunning env check
doc-synceridle
standup-prep8:45am schedule
// ROLE ACTIVITY (today)
Dev0 assists
QE0 assists
DevOps0 assists
Security0 assists
PO/BA0 assists