Dify Python Code Node

Own the Python source and its input/output contract. The companion dify-dsl skill owns workflow graph wiring, YAML serialization, and layout.

Scope Boundary

• Apply this skill to every new or materially revised Python Code node. Do not rely on a subjective "complex code" threshold.
• Reusing an unchanged registered script does not require rewriting it.
• Prefer native Dify nodes such as Template Transform for trivial templating instead of creating unnecessary Code nodes.
• Do not edit the whole DSL. Return a source artifact and a contract that dify-dsl can bind into the graph.

Source Storage

1. Follow repository instructions when they define a Code source registry.
2. When the target DSL is under dev-dsl/, store source under docs/dify-code-nodes/. This repository mode is strict: every Code node must include Code source: <path> in data.desc and the embedded code must match the file.
3. Otherwise, store reusable sources under dify-code-nodes/ at the project root unless the user requests chat-only code.
4. Use descriptive snake_case filenames and group workflow-specific scripts in a subdirectory.

Implementation Requirements

• Define a top-level main(...) function and return a dictionary.
• Keep input parameter names identical to Dify Code node input names.
• Keep every statically returned key represented in the output contract.
• Prefer Python standard library dependencies unless the target sandbox is known to include more.
• Do not use CLI arguments, stdin/stdout, shell commands, local file I/O, or environment variables inside Code node code.
• Do not make network calls unless the user explicitly requires them and the target sandbox behavior is known.
• Use deterministic parsing and transformation logic. Return branchable validation fields for expected input errors instead of raising.
• Keep deep or installation-specific payloads shallow when returned as Object values. Add a JSON String twin only when a real downstream consumer needs the full payload.
• Do not hardcode a universal maximum array length. Instance limits are configurable; use a target-specific limit only when the target runtime or repository policy establishes one.
• Include concise Dify input variables: and Dify output variables: comments before main(...).

Workflow

1. Read the downstream requirements supplied by dify-dsl or the user.
2. Define the smallest useful input/output contract.
3. Write or update the canonical .py source.
4. Add focused tests for parsing, invalid input, boundary behavior, and output shape when a test structure exists.
5. Run Python syntax/tests without executing unavailable Dify services.
6. Return the handoff contract described in references/handoff-contract.md.

Final Handoff

Always report:

• canonical source path;
• inputs with Dify type, required status, and expected selector/value source;
• outputs with Dify type and downstream purpose;
• branch and iteration fields, when present;
• tests executed and any target-runtime assumptions.

The caller must embed the exact source text into data.code; do not provide a second independently edited inline implementation.