agent-runtime-capability-audit

Agent runtime capability gaps occur when skills, CLI tools, or network reachability are missing despite a correct manifest. You can diagnose these gaps by inspecting image and PVC layers with kubectl and curl to validate skill presence, CLI availability, and service access.

Hermes agent task failures often stem from runtime capability gaps rather than config issues. Missing skills, absent CLI tools like curl or wget, or blocked internal services in Kubernetes PVC contexts prevent task execution even when the manifest appears fully valid.

Audit network reachability in Kubernetes agent deployments by using kubectl to inspect the environment and curl or Python fallbacks to test internal service endpoints. This validates whether restricted access across image and PVC layers is blocking agent task execution.

Yes, you need kubectl and curl or Python fallbacks to troubleshoot missing CLI tools in agent PVC layers. These standard debugging tools validate each audit layer, confirm tool availability, and surface concrete fixes for absent curl or wget binaries.

The best way to troubleshoot a Hermes agent with blocked internal service access is a layered runtime audit. Inspect skills presence, verify CLI tool availability, and test network reachability using kubectl and curl across image and PVC layers to identify restrictions.

You should not use a runtime capability audit when the agent manifest, secrets, or configuration themselves are incorrect. This diagnostic specifically targets runtime gaps in skills, CLI tools, and network reachability, not configuration validation or syntax errors.