write-seed

name	write-seed
description	Use when creating golden seed data for a feature

Profile gate: Only for full pipeline. Skip for standard and lite. Check: ptsd context --agent

Instructions

1. Create seed.yaml with feature field and files list.
2. Include at least one happy-path data file.
3. Include edge-case data: empty collections, boundary values, invalid inputs.
4. Use realistic data -- not "test" or "foo".
5. Every file referenced in seed.yaml must exist on disk.
6. Formats: JSON, YAML, or CSV depending on what the feature consumes.

Common Mistakes

• Using placeholder data ("test", "foo", "bar") instead of realistic values.
• Listing files in seed.yaml that do not exist on disk.
• Only covering the happy path -- missing empty, boundary, and invalid cases.
• Creating seed data that does not match the format the feature actually consumes.
• Forgetting the feature field in seed.yaml -- ptsd cannot link it without this.