Overview

Duration	30 minutes
Level	Intermediate
Prerequisites	At least one of Lab 03, Lab 04, or Lab 05

Learning Objectives

By the end of this lab, you will be able to:

• Understand the SARIF v2.1.0 JSON structure and its key fields
• Navigate SARIF findings using the SARIF Viewer extension in VS Code
• Explain the category prefix system used by the Agentic Accelerator Framework
• Connect SARIF output to the GitHub Security tab ingestion pipeline

Exercises

Exercise 6.1: Examine Raw SARIF

Open the SARIF file and explore its JSON structure to understand how agent findings are stored.

1. In VS Code, open validation-results.sarif from the repository root. The file opens as a standard JSON document.
2. Locate the top-level $schema field. This points to the SARIF v2.1.0 JSON schema and tells processors which specification version to validate against.
3. Find the version field. It should read "2.1.0".

4. Expand the runs[] array. Each run represents one tool execution. Inside a run, identify these fields:

   Field	Purpose
   tool.driver.name	Name of the agent or scanner that produced the results
   tool.driver.rules[]	Rule definitions with unique ruleId values per finding type
   results[]	Array of individual findings
   automationDetails.id	Category string used for grouping (for example, security/)

5. Inside a single result entry, locate:

   Field	Purpose
   ruleId	Unique identifier for the rule that triggered the finding
   level	Severity level (error, warning, or note)
   message.text	Human-readable description of the finding
   locations[]	File path and line number where the issue was detected
   partialFingerprints	Stable hashes used for deduplication across multiple runs

6. Count how many runs exist in the file and note which tools produced them.

Exercise 6.2: Use SARIF Viewer Extension

The SARIF Viewer extension provides a graphical interface for navigating findings without reading raw JSON.

1. Right-click validation-results.sarif in the VS Code Explorer panel.
2. Select Open with SARIF Viewer from the context menu. If the option does not appear, confirm the SARIF Viewer extension is installed (see Lab 00 setup).
3. The viewer displays a findings tree grouped by tool and severity. Expand a tool node to see its individual findings.

1. Click any finding in the tree. The viewer opens the referenced source file and highlights the exact line where the issue was detected.
2. Compare the source location shown in the viewer to the locations[] field you examined in Exercise 6.1. Both should point to the same file and line number.

Exercise 6.3: SARIF Category Prefixes

The Agentic Accelerator Framework uses category prefixes in the automationDetails.id field to organize findings by domain.

1. Review the category prefix system:

   Prefix	Domain	Example Agents
   security/	Security scanning	SecurityReviewerAgent, IaC Security Agent, Supply Chain Agent
   accessibility-scan/	Accessibility	A11Y Detector, A11Y Resolver
   code-quality/coverage/	Code quality	Code Quality Detector, Test Generator
   finops-finding/v1	FinOps governance	Cost Analysis Agent, Cost Optimizer Agent

2. Understand the severity mapping between SARIF levels and framework classification:

   SARIF Level	Framework Severity	Action Required
   error	CRITICAL or HIGH	Must remediate before merge
   warning	MEDIUM	Address in current sprint
   note	LOW	Track for future improvement

3. Return to the raw SARIF file and search for automationDetails entries. Identify which category prefix each run uses.
4. For security findings, the framework maps to CWE IDs (for example, CWE-79 for XSS) and OWASP Top 10 categories. For accessibility findings, the mapping references WCAG 2.2 success criteria.

Exercise 6.4: How GitHub Ingests SARIF

This exercise explains the pipeline from SARIF file to GitHub Security tab. Labs 07 and 08 will walk you through the process hands-on.

1. Understand the SARIF upload flow:

   Agent produces findings
        ↓
   Results written as SARIF v2.1.0 JSON
        ↓
   GitHub Actions workflow runs upload-sarif action
        ↓
   GitHub Code Scanning processes the SARIF file
        ↓
   Findings appear in Security → Code scanning alerts
   

2. The github/codeql-action/upload-sarif@v4 action in each workflow handles the upload. The category input on the upload step matches the automationDetails.id prefix so GitHub can group alerts by domain.
3. Once uploaded, GitHub deduplicates findings using partialFingerprints. A finding that already exists from a previous run will not create a duplicate alert.
4. In Lab 07, you will enable GitHub Actions and trigger these workflows with a pull request. In Lab 08, you will explore the uploaded results in the Security tab.

Verification Checkpoint

Before proceeding, verify:

• You can identify the five key SARIF fields: $schema, version, runs[], results[], and partialFingerprints
• You opened the SARIF file in the SARIF Viewer extension and navigated to a source location
• You can explain the four category prefixes and which domain each represents
• You understand the severity mapping from SARIF levels to framework classification
• You can describe the flow from SARIF file to GitHub Security tab

Next Steps

Proceed to Lab 07 to enable GitHub Actions workflows and trigger them with a pull request.