Artifacts¶
Artifacts are physical or runtime outputs produced by pipeline nodes.
Examples:
- downloaded CSV file
- transformed dataset
- generated JSON document
- output file
- run result metadata
Artifacts are not generic variables¶
Avoid treating artifacts as vague placeholders. A value like this is ambiguous:
It can be unclear whether that means:
- prior node output
- latest artifact
- first upstream artifact
- file content
- scalar value
- query string
- branch-specific output
Recommended direction¶
Use artifacts for outputs.
Use named runtime values for dynamic field substitution.