fix: update workload status when health check fails for remote MCP servers

[fredericlefeurmou]

Closes #3079

Problem

When a remote MCP server becomes unreachable (e.g., Datadog MCP), the health check fails and the proxy is stopped. However, the workload status is not updated, causing thv list to report the server as "running" even though nothing is listening on the port.

This creates a "ghost status" problem where:

• thv list shows the server as "running"
• Cursor shows the MCP as enabled
• lsof -i :PORT shows nothing listening
• Cursor gets stuck on "Loading tools..."

Solution

Add a callback mechanism that notifies the runner when a health check fails, allowing it to update the workload status to unhealthy.

Changes

• Add HealthCheckFailedCallback type to transport types
• Add onHealthCheckFailed callback to TransparentProxy
• Pass callback from HTTPTransport to TransparentProxy
• Runner sets callback for remote servers to update status manager
• Call callback in monitorHealth() before stopping proxy

Result

When a remote MCP server dies:

1. Health check detects the failure
2. Callback notifies the runner
3. Runner updates status to unhealthy
4. thv list shows accurate status

[Copilot]

Pull request overview

This PR fixes a "ghost status" issue where remote MCP servers that become unreachable continue to show as "running" in thv list, even though the health check has failed and the proxy has stopped. The solution adds a callback mechanism that allows the proxy to notify the runner when health checks fail, enabling proper workload status updates.

Key Changes:

• Added HealthCheckFailedCallback type to allow notification of health check failures
• Integrated callback into the transparent proxy health monitoring system
• Connected the callback to the status manager in the runner for remote servers

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
pkg/transport/types/transport.go	Defines HealthCheckFailedCallback type and adds OnHealthCheckFailed field to Config struct
pkg/transport/proxy/transparent/transparent_proxy.go	Adds onHealthCheckFailed callback parameter to constructor and invokes it when health check fails
pkg/transport/proxy/transparent/transparent_test.go	Updates all test calls to NewTransparentProxy with nil callback parameter
pkg/transport/http.go	Adds SetOnHealthCheckFailed setter method and passes callback to TransparentProxy
pkg/runner/runner.go	Sets health check failure callback for remote servers to update workload status to unhealthy
cmd/thv/app/proxy.go	Passes nil callback for local proxy command (not needed for local proxies)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codecov]

Codecov Report

❌ Patch coverage is 39.28571% with 17 lines in your changes missing coverage. Please review.
✅ Project coverage is 56.80%. Comparing base (0a7bacf) to head (aea9a24).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
pkg/runner/runner.go	0.00%	12 Missing ⚠️
pkg/transport/http.go	0.00%	3 Missing ⚠️
...g/transport/proxy/transparent/transparent_proxy.go	84.61%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3077      +/-   ##
==========================================
- Coverage   56.82%   56.80%   -0.03%     
==========================================
  Files         335      335              
  Lines       33474    33492      +18     
==========================================
+ Hits        19022    19025       +3     
- Misses      12868    12883      +15     
  Partials     1584     1584              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.