Skip to main content
GitHub Agentic Workflows
DocsWorkflow examplesBlog
Get started
  • Docs
  • Workflow examples
  • Blog

Documentation

  • Blog
  • Introduction
    • What are Agentic Workflows?
    • How Agentic Workflows Work
    • Security Architecture
  • Setup
    • Quick Start
    • Creating New Workflows
    • Working with Workflows
    • Creation Wizard
    • CLI Commands
  • AI Engines
    • GitHub Copilot
    • Claude Code
    • OpenAI Codex
    • Google Gemini
    • Agy (experimental)
    • Pi
  • Guides
    • GitHub Actions Primer
    • Using MCPs
    • Network Configuration
    • Deploying a Work Queue
    • Agentic Automation Workshop
    • Using at Scale in Organizations
  • Gallery
    • Gallery Overview
    • Issue Triage
    • Pull Request Review
    • Documentation Maintenance
    • Repository Maintenance
    • CI Failure Investigation
    • Code Improvement
    • Dependency Analysis
    • Metrics and Analytics
    • Repository Reporting
    • Security Review
    • Multi-Repository Examples
    • Triage from Side Repo
    • Code Quality Monitoring
    • Feature Synchronization
    • Cross-Repository Issue Tracking
    • Dependabot Rollout
  • Design Patterns
    • BatchOps
    • CentralRepoOps
    • ChatOps
    • Daily Report Portfolio
    • DeterministicOps
    • DispatchOps
    • IssueOps
    • LabelOps
    • Linter Factory
    • MemoryOps
    • MonitorOps
    • FeatureOps
    • MultiRepoOps
    • OrchestratorOps
    • ProjectOps
    • ResearchPlanAssignOps
    • SpecOps
    • WorkQueueOps
  • Practices
    • Measuring Impact
    • Safe Rollout in Organizations
    • Sharing Workflows in Organizations
  • Reference
    • AI Engines
    • AI Engines (Custom)
    • AI Engines (Azure BYOK)
    • Authentication
    • Authentication (Projects)
    • Compilation Process
    • Configuration Governance
    • Cost Management
    • Cost Management (Auditing Workflows)
    • Cost Management (Billing)
    • Cost Management (Rate Limiting)
    • Cost Management (Model Tables)
    • Enterprise Configuration
    • Enterprise Configuration (Compiler Controls)
    • Environment Variables
    • FAQ
    • GH-AW Agent
    • GH-AW as MCP Server
    • GitHub (Checkout)
    • GitHub (Read Tools)
    • GitHub (Read Permissions)
    • GitHub (Integrity Filtering)
    • GitHub (Cross-Repository)
    • GitHub (Fork Support)
    • GitHub (Artifacts)
    • Glossary
    • Imports
    • Imports (APM)
    • Imports (Copilot Agent Files)
    • Imports (Copilot Inline Sub-Agents)
    • Imports (Dependabot)
    • Outcomes
    • OpenTelemetry
    • OpenTelemetry Attributes
    • Package Manifest
    • Releases and Versioning
    • Safe Outputs
    • Safe Outputs (Copilot Cloud Agent)
    • Safe Outputs (Custom)
    • Safe Outputs (Ephemerals)
    • Safe Outputs (Footers)
    • Safe Outputs (Pull Requests)
    • Safe Outputs (Staged Mode)
    • Safe Outputs (Triggering CI)
    • Sandbox
    • Sandbox (Agent Runtimes)
    • Sandbox (MCP Gateway)
    • Sandbox (Network Access)
    • Self-Hosted Runners
    • Self-Hosted Runners (ARC DinD)
    • Threat Detection
    • Tools
    • Tools (Cache Memory)
    • Tools (Repo Memory)
    • Tools (Playwright)
    • Tools (Serena)
    • Tools (Web Search)
    • Work Queues
    • Workflows
    • Workflows (Frontmatter)
    • Workflows (Markdown)
    • Workflows (Triggers)
    • Workflows (Schedule Syntax)
    • Workflows (Command Triggers)
    • Workflows (Concurrency)
    • Workflows (Custom Steps and Jobs)
    • Workflows (Feature Flags)
    • Workflows (MCP Scripts)
    • Workflows (Templating)
    • Workflows (Full)
  • Specs
    • Package Management (Spec)
    • Forecast Command Specification
    • Frontmatter Hash Specification
    • Fuzzy Schedule Time Syntax Specification
    • Graders Specification
    • MCP Scripts Specification
    • Model Alias Format Specification
    • Safe Outputs MCP Gateway Specification
    • AI Credits Specification
    • Checkout Behavior Specification
    • Copilot SDK Driver Specification
    • Unified Agent Session Specification
    • Safe Update Specification
    • Bash Command Parser Specification
    • Git-backed Ledger Specification
    • Work Queue Specification
  • Contributing
    • Error Message Style Guide
  • Troubleshooting
    • Error Reference
    • Common Issues
    • Debugging Workflows
    • Debugging GHE Cloud with Data Residency
  • Experimental
    • A/B Experiments
    • A/B Experiments (Spec)
    • AWF Reflect Route
    • CorrectionOps
    • Drive Memory (Private Preview)
    • Private Repository Enclaves
    • Graders
    • Monitoring with Projects
    • QMD Documentation Search
    • Safe-Outputs Samples
    • TrialOps
    • WASM Compilation
  • Agent Factory
  • Editors
  • About
GitHub
Get started
On this page
  • Overview
  • Declaring experiments
    • Bare-array form
    • Rich object form
  • Using variants in the prompt
  • Common experiment ideas
    • Try different prompt styles
    • Try different models
    • Try using a sub-agent
    • Try different subskills
  • Statistical balancing
  • Storage Configuration
  • Accessing assignments downstream
  • Analyzing results
    • Analyze a grader-backed metric
    • Analyze an eval-backed metric
    • Deterministic decisions
      • Human-readable output
      • Reason codes
    • Filtering audit results by variant
    • Step summary
  • Frontmatter reference
    • Bare-array form
    • Object form fields
  • Continual experiment ramps

On this page

  • Overview