Skip to content
Go to Boltz API
Protein
Design

Run

Custom CLI workflow command for `boltz-api protein:design run`.

Start a pipeline run, wait for completion, and download results. Prefer --input for full payloads; top-level flags remain available as overrides.

This is CLI workflow behavior, not a standalone REST endpoint. It composes generated API commands from this resource to submit work, wait for completion, and persist local results.

Terminal window
boltz-api protein:design run
OptionDescription
--input [ --input ]Structured request input. Merges fields into the top-level request body. Prefer --input for complete structured payloads or @json://... files. Top-level body flags can still override individual fields. Keep --idempotency-key and --workspace-id top-level; if also present in --input, the top-level flag wins.
--binder-specification [ --binder-specification ]Binder specification for protein design. Use no_template for sequence-defined binders, structure_template for uploaded binder structures, or boltz_curated for Boltz-managed nanobody and antibody defaults. You can set this directly, or prefer --input for the full payload. If both are set, this flag overrides the matching field from --input.
--num-proteinsNumber of protein designs to generate. Must be between 10 and 1,000,000. You can set this directly, or prefer --input for the full payload. If both are set, this flag overrides the matching field from --input.
--target [ --target ]Target specification (structure template or template-free) You can set this directly, or prefer --input for the full payload. If both are set, this flag overrides the matching field from --input.
--idempotency-keyClient-provided key to prevent duplicate submissions on retries. Keep this as a top-level flag even when using --input. If also present in --input, this top-level flag wins.
--workspace-idTarget workspace ID (admin keys only; ignored for workspace keys). Keep this as a top-level flag even when using --input. If also present in --input, this top-level flag wins.
--name stringLocal run directory name under --root-dir
--run-dir stringExplicit local run directory path
--root-dir stringRoot directory for generated local run directories (default: "boltz-experiments")
--poll-interval-seconds floatPolling interval while waiting for remote results (default: 5)
--progress-format jsonlFormat for progress logs written to stderr (jsonl or text). jsonl is the default agent-friendly format. (default: "jsonl")
--verbose, -vPrint text progress logs to stderr when --progress-format text is selected (default: false)
--download-mode everythingPipeline artifact download mode (everything or metadata_only). (default: "everything")
--workers int, --num-workers int, --num_workers intNumber of concurrent pipeline result archive downloads (default: 32)
--help, -hshow help