Creates a connector tunnel, attaches a published hostname or a private network route to it, diagnoses
its health, and retires it, following the platform's own documented procedure
(references/<vendor>/procedure.md).
When to use this skill
- An internal web or TCP service needs to be reachable from the Internet or from authorized users
without an inbound firewall rule: create the tunnel and publish a hostname.
- A private IP range or private hostname needs to be reachable through a connector installed inside
that network: create the tunnel and add a route.
- A tunnel shows as inactive, down, or degraded, or users get an error for a published application:
diagnose it.
- A tunnel, its routes, or its connectors need to be removed (retired).
- Not for site-to-site IPsec or GRE tunnels between routers (see
vpn-tunnel-management), for the
access policy that decides who may reach the application (see
zero-trust-access-policy-management), or for the DNS records of a zone beyond the one the
platform's procedure requires (see dns-record-management).
Universal concepts (apply regardless of platform)
- A connector tunnel is outbound-only. The connector runs inside the private network and opens
connections out to the provider, so exposing a service does not require an inbound rule. Where
documented, the tunnel object and the connector process are separate: creating the tunnel does not
make it carry traffic until a connector runs with the tunnel's credential.
- The tunnel credential is a secret. Anyone holding it can run the tunnel. Handle it as a secret
and do not echo it into logs or reports.
- Where the platform documents it, tunnel health describes the connector's link to the provider
only. A tunnel can read healthy while the application behind it fails, so check the internal
service separately.
- Where the platform documents it, routing is a separate record from the tunnel. Reaching a
service needs a route or published hostname attached to the tunnel.
- Only one implementation is listed in
references/INDEX.md; each claim above holds only where the
platform documents it.
Procedure
- Identify the target platform from the request or the
platform argument. If it is not in
references/INDEX.md, stop: this skill does not cover it and does not improvise a procedure. - Open
references/<platform>/procedure.md and follow it exactly, including its confirmation
gates. It holds the full steps, error handling, patterns, limitations, tools, and verification
checklist. - Report the outcome using that procedure's own status vocabulary, not a normalized one.
Known limitations
- Platform coverage is exactly the platforms in
references/INDEX.md. - Does not cover the policy that governs who may use a published application or private route, the
provider's load balancing, or the internal service's own configuration.
- Does not cover site-to-site IPsec or GRE tunnels; those are routed to
vpn-tunnel-management.
Contributing to this skill
SKILL.md itself should only ever change for two reasons: a bug in the universal concepts or router
logic above, or adding a pointer to a newly supported platform. Platform-specific changes — a new
error, a corrected field, a changed path — belong in that platform's own references/<platform>/
files, never here.
Tools
Each platform assumes its own authenticated client; see references/INDEX.md for which, and that
platform's procedure.md for the operations.
Verification checklist
- [ ] Target platform confirmed against
references/INDEX.md - [ ] That platform's
procedure.md was followed, including its confirmation gates - [ ] The tunnel credential was not written into any output
- [ ] Outcome reported using the platform procedure's own status vocabulary