BuildRunner

File: tckit/ports/builder.py Purpose: Build, deploy, and runtime control with structured results.

Method	Returns
build(project_path, *, plc_name=None)	BuildResult
deploy(target_ams_id, *, plc_name=None)	Result
start_runtime(target_ams_id)	Result
get_status()	BuildStatus

BuildResult carries each diagnostic as {file, line, message, severity} — never a raw log blob. Never deploy without a successful build first.

build and deploy accept an optional plc_name to scope the operation to a single PLC project on multi-project solutions; start_runtime is target-wide. The bridge follows the standard fallback chain (per-call name → PLC_PROJECT_NAME env → auto-resolve → ambiguous error).

Why this shape

Tool outputs are context. A 2,000-line build log forces the model to scan, regex, and guess; a list of structured diagnostics lets it act. This is the high-signal tool result principle — return what the agent will use, not everything the underlying tool emitted.