Generating Workflows

Workflow Generation

Workflow pattern generation and customization skill with mandatory schema validation against schemas/workflow.schema.json

Generates workflow configurations and documentation based on methodology and trigger sources, with mandatory schema validation against schemas/workflow.schema.json.

Canonical Schema Reference

Workflow frontmatter MUST validate against schemas/workflow.schema.json. Required fields:

| Field | Type | Constraints | |-||-| | name | string | Min 3 chars | | description | string | Min 20 chars | | version | string | Semantic version | | type | enum | ONLY: supervisor, sequential, parallel, hierarchical, iterative, pipeline | | domain | enum | ONLY: ai-ml, agile, blockchain, dotnet, java, operations, python, quality, sap, trading, typescript, universal | | steps | array | Min 2 items, each with required name and description | | agents | array | Min 1 item | | blueprints | array | Min 1 item (use ["none"] if not applicable) |

Domain mapping: data-science → python or ai-ml, automation → operations, general → universal

Validation Checklist

• All required frontmatter fields present
• type and domain use allowed enum values
• At least 2 steps, each with name and description
• At least 1 agent listed
• blueprints has min 1 item (use ["none"] when no blueprint)
• **Version:** x.x.x included in file body (CI validation)
• ## Overview and ## Trigger Conditions sections present and populated

Common Violations

Process

1. Review the task requirements.
2. Apply the skill's methodology.
3. Validate the output against the defined criteria.

Step 1: Parse Workflow Requirements

• Identify methodology (Agile, Kanban, etc.)
• List trigger sources (Jira, Confluence, etc.)
• List required output artifacts

Step 2: Select Workflow Patterns

Match triggers to workflow patterns:

| Trigger | Patterns | ||-| | Jira | bugfix-workflow | | Confluence | feature-workflow | | GitHub Issue | bugfix-workflow | | GitHub PR | code-review | | Manual | code-templates |

Step 3: Configure MCP Servers

For each trigger, configure required MCP server:

| Trigger | MCP Server | Authentication | |||-| | Jira | atlassian | OAuth | | Confluence | atlassian | OAuth | | GitHub | deepwiki | None | | SAP Docs | sap-documentation | None |

Generate MCP configuration:

mcpServers:
  atlassian:
    url: "https://mcp.atlassian.com/v1/sse"
    headers: {}

Step 4: Generate Workflow Files

For each selected pattern, generate documentation:

• {directories.workflows}/{pattern_name}.md
• REQUIRED: Include **Version:** x.x.x in the file (typically in the Overview/metadata block). This is validated by CI tests and will fail the build if missing.
• REQUIRED: Include ## Overview and ## Trigger Conditions sections.
• Include step-by-step process.
• Include artifact outputs.
• Run verify_structures.py to confirm structural integrity.

Step 5: Output Configuration

workflows:
  methodology: "{METHODOLOGY}"
  patterns: ["bugfix-workflow", "feature-workflow"]
  triggers:
    - type: "jira"
      pattern: "{PROJECT_KEY}-{NUMBER}"
    - type: "confluence"
      pagePattern: "Page ID or URL"
  mcpServers: [...]

Step 6: Post-Creation Sync

When adding workflows to the factory:

python scripts/validation/validate_readme_structure.py --update
python scripts/validation/verify_structures.py

Best Practices

Best Practices

• Match workflow patterns to actual team processes: Don't force-fit patterns; customize workflows to match how the team actually works.
• Validate MCP server availability before workflow creation: Check that required MCP servers are accessible and properly configured.
• Design workflows with clear trigger conditions: Define specific, testable conditions for when workflows activate.
• Include manual fallback procedures: Always document what to do when automated steps fail.
• Test workflow execution paths: Validate that all decision branches and escalation paths work correctly.
• Document workflow dependencies: Clearly list all external systems, credentials, and prerequisites.

Important Rules

1. Use {directories.XXX} path variables — NEVER hardcode directory paths like workflows/ or knowledge/ in generated workflow content. See {directories.config}/settings.json for the full mapping.
2. Lowercase kebab-case filenames — All workflow files must use lowercase kebab-case (e.g. feature-development.md).
3. Validate against schema — Ensure frontmatter validates against schemas/workflow.schema.json.

Fallback Procedures

• If trigger not supported: Ask user for manual workflow design
• If MCP server unavailable: Document manual fallback procedure

References

• {directories.knowledge}/workflow-patterns.json
• {directories.knowledge}/mcp-servers-catalog.json
• {directories.knowledge}/knowledge-cross-reference.json
• schemas/workflow.schema.json

When to Use

This skill should be used when strict adherence to the defined process is required.

Prerequisites

• Basic understanding of the agent factory context.
• Access to the necessary tools and resources.