[SizzleUnrlsd]

Contract

ctrace deprecated its --async for -j/--jobs (CoreTrace/coretrace#158, 2026-10-01): tools run on one worker per core by default, --async is a deprecated spelling of that default, and ctrace runs this analyzer with --check --format sarif <root>, waiting for the result. Rather than adding an --async mode the orchestrator has just retired, the analyzer follows the same model:

• -j N is --jobs N; --jobs 0 analyses in one process per core the process may run on (os.process_cpu_count), as runtime.jobs of ctrace; the default stays 1, so a run without the option is unchanged; a negative value is an error.
• The command always waits for the complete report. Report, finding order, --fail-on, cache, baseline and SBOM/VEX do not depend on the number of processes; worker processes are bounded and cleaned up by the existing process pool.
• There is no --async option, and no detached job.

engine.analyze_project keeps its jobs >= 1 contract; the command line resolves 0.

Verification

New tests in tests/test_jobs_option.py (red before the change): -j equals --jobs, -j 0 uses one process per core and yields the same report, the default stays one, negative values are rejected, --async does not exist, the help names -j and 0. No existing test changed. Full suite green on 3.14 (apart from the local test_packaging environment issue) and 3.11; ruff, mypy clean; regression snapshots unchanged; mitmproxy: -j 0 (8 cores) gives a report identical to -j 1.

Closes #205