Updating an import, and exporting
What an OpenSpec or Spec Kit import brings in, how a Spec Kit import takes the features added since, and how to export the spec back as files for your repository, or in other formats.
Importing a repository
To start a project from a spec your team keeps in a repository, open Start from a template or import on the home screen and choose OpenSpec or Spec Kit. Paste a public GitHub repository, or, for a private one, upload a .zip or choose its folder: only the spec files are sent. Import from OpenSpec and Import from Spec Kit show what you get, and let you try it before signing up.
| OpenSpec | Spec Kit | |
|---|---|---|
| What is read | openspec/specs/: each spec.md. Not openspec/changes/. | specs/: each feature's spec.md and tasks.md. |
| Requirements | Each ### Requirement:, with its scenarios as acceptance criteria. Each spec domain becomes a subsystem. | Each FR-, NFR- and SC- item, word for word. |
| Stories | Drafted by Entalpa from your specs, with the stakeholders and the open questions. | Each user story, as written. Stakeholders come from the stories' roles. |
| Status | Draft. | From the ticks in tasks.md: implemented when all are done, in progress when some are. |
| Priority | From the requirement's keyword: MUST or SHALL is Must, SHOULD is Should, MAY is Could. Without one, Must. | The same for requirements. A story's P1 is Must, P2 Should, P3 Could. |
| Questions | Drafted with the stories. | Clarifications come in as answered questions, [NEEDS CLARIFICATION] markers as open ones. |
| Then, with AI | Entalpa writes the system-level requirements above your specs and links them to the stories. | Entalpa folds the features into one backlog, oldest first: a later feature revises the story it changes instead of adding a copy. |
Reading the repository is free; the AI step that follows uses credits. An import brings in up to 300 requirements, and says so when a spec has more. Every imported project starts with one saved version, Initial version.
Updating a Spec Kit import
A project imported from a public GitHub repository with Spec Kit has Import new Spec Kit features in its menu. It reads the same repository, branch and folder again, brings in the feature folders added since, as written, and folds them into the backlog. Features it imported before are not read again; if their files changed, it says which.
- Folding stops? What was folded is saved. Continue from Import new Spec Kit features, or from Fold the features on the Overview.
- Imported from a .zip or a folder? There is no repository to read again. Uploading the newer specs starts a separate project.
- OpenSpec? An import cannot be updated: importing the repository again makes a new project. To keep a project in step as OpenSpec changes are archived, use the entalpa-openspec skill.
Exporting
Everything goes out from Export in the project menu. For your repository, OpenSpec writes openspec/specs/<subsystem>/spec.md for each subsystem and openspec/specs/system/spec.md for the system-level requirements, meant to replace what openspec/specs/ holds; openspec/changes/ is not touched. Spec Kit writes the whole current spec as one feature, specs/<NNN>-current-system/spec.md, numbered after your imported features. Both download as a zip laid out to unpack at the repository's root, and both leave out rejected and superseded requirements.
| In the Export menu | What you get |
|---|---|
| Requirements: ReqIF, CSV, Markdown | The requirements, for a requirements tool, a spreadsheet or a document. |
| Stories: JSON, Markdown | The stakeholders and their user stories. |
| Spec files for your repository: OpenSpec, Spec Kit | The spec as files, as above. |
| Description | A description of the project written for one stakeholder, with AI. |
| Whole project: Portable JSON, Portable YAML | Everything, to import again: see The project file. |
| PRD / BRD document: PDF, Markdown | A document for stakeholders. Its narrative is written with AI, and uses credits. |