Connect the records that answer the question.
Helio is not another place to write documentation.
What can go into the graph
GitHub
AvailableReads issues, pull requests and commit metadata for one owner and repository. Used as a recency signal: who touched the code, and when. Read-only; Helio never writes back.
Services and dependencies
AvailableServices, owning teams and HAS_DEPENDENT edges are loaded from structured records. The demo ships with a 20-service company so the graph can be evaluated before a real import.
On-call rotations
Available via importRotation membership and the active weekly on-call are separate facts. Helio stores both, so "on the rotation" is not confused with "paging now".
Decision records
Available via importADRs and architecture notes are stored with dates and SUPERSEDES edges. A newer decision can retire an older document without deleting it.
Helio API
Available on paid plansCreate an API key in the account and query the same graph from your own tooling. Keys are scoped to the workspace and can be revoked.
PagerDuty, Opsgenie, Jira, Confluence
Not native yetThere is no one-click connector for these today. Their records can be imported through the graph loader, and Enterprise onboarding can scope a connector if that is the blocker.
Source systems stay the source of truth
Helio stores relationships and dates, not copies of every document. A GitHub issue remains in GitHub; the graph records that it exists, who is involved and when it changed. A decision record remains in the repository; the graph records that D-19 supersedes the March document.
That keeps the import small and makes staleness visible. If a source has not been synced, the date on the edge shows it. The answer can still use the edge, but it cannot pretend the edge is fresh.
Integration questions
- Does the GitHub connection need write access?
- No. It reads issues, pull requests and commit metadata for the repository named in the deployment. It does not comment, merge, label or change issues.
- Can we import our own services and dependencies?
- Yes. Services, teams, people, dependencies, on-call rotations and decisions are loaded as structured records. The loader is the same one used to build the demo company, so the shape is documented in the repository.
- Is there a real-time PagerDuty sync?
- Not today. On-call data is imported into the graph. If you need a native connector, say so during an Enterprise conversation; we would rather scope it than claim it exists.
- What does the API return?
- The same answer and trail the agent shows, including the graph path, dates and any missing edge. API access is available on Professional and Enterprise plans.
Have a source that is not listed?
Tell us what holds the record and how often it changes. We will say whether it can be imported today or whether it needs a connector.