AddThisFeature

Integration Connection Health

Show at a glance whether each connected service actually still works.

moderate Integrations

What it adds

A health state per integration connection that separates expired credentials, missing permissions, and provider outages.

What your agent is told to do

6
  1. 1

    Give every connection an explicit health state: healthy, needs reconnection, missing permission, provider unavailable, or never used.

  2. 2

    Derive health from the last successful call, not from whether a token row exists. A stored token proves nothing — providers revoke without telling you.

  3. 3

    Record the classified reason whenever a call fails, so the state has an explanation attached rather than being recomputed from guesswork.

  4. 4

    Offer a reconnect action that repairs the existing connection in place, keeping its ID, settings, and mappings. Do NOT create a second connection on reconnect — that orphans everything attached to the first one.

  5. 5

    Show the user a plain sentence about what to do. Never surface the provider's raw error body, which routinely contains tokens, internal URLs, and account identifiers.

  6. 6

    Per-run sync freshness, record counts, and the manual sync button are owned by External Sync Status; link to it rather than duplicating it. Refresh-token mechanics are owned by OAuth Token Refresh Recovery.

Edge cases it handles

6
  • A provider outage is not a broken connection. Do not tell the user to reconnect when the fix is to wait.
  • A scope added to the app after a user connected leaves old connections technically valid but functionally incomplete. Detect the missing scope and ask for incremental consent, not a full teardown.
  • A connection that has never been used has no last-success timestamp. Show 'not yet used' rather than an alarming red state.
  • Rate limiting looks like failure. Treat it as a distinct, temporary state and do not degrade health for it.
  • Health checks must not hammer the provider. Judge from real traffic, and probe on a schedule slow enough not to consume the user's quota.
  • The person who sees the broken connection may not be the one who can fix it. Say who owns the connection and notify them.

Definition of done

8
  • Every connection displays one of the defined health states.
  • Expired credentials, missing scopes, and provider outages are distinguishable in the UI.
  • Health is computed from the last successful call, not token presence.
  • Reconnecting preserves the connection's ID and configuration.
  • No provider error body, token, or internal URL is shown to the user.
  • Health checking does not measurably consume the user's provider rate limit.
  • The feature matches the existing design system.
  • No existing functionality is broken.

Related features

How it works

  1. 1

    Copy the link

    Grab the Markdown instruction URL for this feature.

  2. 2

    Give it to your AI

    Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.

  3. 3

    It inspects, then implements

    Your agent reads your existing app first, then adds the feature to fit it.

Works with your stack

These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.

Need it tighter than that? Customize the feature and tell it exactly what you're running.