Run Evals from the operator CLI
`platphormctl` is a first-class Evals client for discovery, MCP validation, policy inspection, dry-run harnesses, and evidence-producing release checks. Public examples are documented here; protected execution still requires PLATPHORM_API_KEY.
site-inspect-evals
public-safeInspect Evals public route, policy, discovery, and health surfaces.
platphormctl site inspect evals
mcp-validate-evals
public-safeValidate Evals MCP JSON-RPC introspection and tool schema metadata.
platphormctl mcp validate evals
policy-inspect-evals
public-safeInspect Evals agent, AI, trust, security, and robots policies.
platphormctl policy inspect evals
evals-list
public-safeList public-safe Evals suites, templates, gates, and recent run summaries.
platphormctl evals list
evals-run-site-mcp
public-safeRun a site-level evaluation plan for the MCP Hub target.
platphormctl evals run-site mcp
evals-run-mcp-mcp
public-safeRun public-safe MCP introspection checks for MCP Hub.
platphormctl evals run-mcp mcp
grade-tool-health
public-safeGrade the MCP get_health tool using deterministic output checks where possible.
platphormctl evals grade-tool mcp get_health
harness-discovery-full
public-safeRun the full discovery harness with trace propagation.
platphormctl harness run discovery-full --trace
developer-validation-dry-run
dry-runPreview developer validation without protected execution.
platphormctl harness run developer-validation --target https://evals.platphormnews.com --dry-run
spec-evals-browserops-loop
dry-runPreview the Spec to Evals to BrowserOps loop without claiming provider evidence.
platphormctl harness run spec-evals-browserops-loop --dry-run
Evidence rule
This page documents commands; it does not claim they executed. Store real CLI output through a protected runner, `platphormctl` dry-run artifact, or persisted Evals run before using CLI evidence in scorecards or release gates.