Entalpa

For OpenSpec projects

The spec that outlives the change

OpenSpec is the lightest credible way to take one change from intent to merged. Paste your repository and see what the corpus looks like after fifty of them — read-only, no account, and nothing written back to your repo.

Public repositories only, read through the GitHub API. We never write to yours.

What happens when you paste a repository

1

We read specs/, not changes/

One subsystem per domain, one requirement per heading, scenarios kept as acceptance criteria — titles and all, because the title is the part a person wrote.

2

Then the layers the format has nowhere to keep

Stakeholders and stories, inferred from your own requirements and labelled as inferred. And the open questions: the places your spec is silent, which you can check against the file.

3

You read the whole thing, before signing up

Not three teaser rows. Your repository is public, so an enriched view of it withholds nothing you had not already published — and the graph is the only honest way to show what this is for.

Three things the format cannot do for itself

None of these is a criticism of the workflow — propose → apply → archive is a good loop and we are not asking you to stop running it. They are properties of keeping the corpus in markdown, and each is checkable against OpenSpec’s own documentation.

No identity outside the repository

A rename inside openspec/ is handled properly — ## RENAMED Requirements exists, and archive retitles in place. What cannot exist is a reference from outside: the heading is the identity, so a ticket, a test or a commit message has nothing durable to name.

History at file level, not requirement level

Git is your history, and it is a good one. What it cannot give you is a diff keyed to a requirement — theirs is a diff of a three-hundred-line markdown file, where the question you actually ask is “what happened to this one requirement, and why”.

Nothing to query

Three hundred requirements across twenty domain folders, and the tools are grep and memory. Which requirements has nobody traced to a user? Which changed since the release? Those are reads, and a folder of markdown does not answer reads.

OpenSpec stays the source of truth

openspec/specs/ stays authoritative and your repository stays yours. Entalpa reads it and adds what markdown has no room for — identity, a history per requirement, and something to query. Keep running openspec exactly as you do now.