Fakecode To Workflow

Fakecode To Workflow

Translate between loose human fakecode and agent-readable workflow documents. Treat user input as intent, not syntax; do not validate Kotlin, Chinese punctuation, brackets, or indentation.

Route

Choose exactly the reference needed for the user's request:

• Fakecode/pseudocode -> workflow: read fakecode-to-workflow.md.
• Existing workflow -> fakecode/pseudocode: read workflow-to-fakecode.md.
• Update an existing workflow from user fakecode or edits: read update-existing-workflow.md.

When fakecode syntax matters, also read syntax.md. When behavior is unclear or examples would disambiguate the pattern, read examples.md. Before final workflow output or file updates, read and apply workflow-lint.md. For update tasks, also read workflow-to-fakecode.md if you need to normalize the current workflow before applying changes.

Invariants

• Debug logic and closure risks before producing a final workflow.
• Lint generated workflows for agent executability before final output or file updates.
• Final workflows must be agent-readable: explicit actions, inputs, decisions, outputs, loops, failure paths, artifacts, stop conditions, and out-of-scope boundaries.
• Use ASCII Markdown flowcharts only; do not use Mermaid or images.
• Ask the user to confirm unresolved boundaries before final generation or file updates, unless they explicitly asked to apply directly and the risks are already clear.