JSON schemas¶
Machine-readable outputs are versioned as their contracts stabilize.
doctor --json¶
The current beta document contains cli_version, platform, python, ffmpeg, ffprobe, and capabilities. The versioned capability inventory includes full encoder, decoder, filter, muxer, demuxer, protocol, subtitle, and hardware-accelerator facts plus core availability summaries.
probe --json¶
The simplified probe document contains format fields, normalized duration/size/bitrate, tags, stream summaries, dispositions, language, rotation, color/HDR facts, side data, attachments, and chapters. Python callers can request raw=True or call probe_raw() for the lossless FFprobe document.
smoke-test --json¶
The document uses schema_version: "1.0" and reports status, retention behavior, workspace policy, and probe summaries for the synthetic input and generated thumbnail.
Plans, results, profiles, and preflight¶
ExecutionPlanusesschema_version: "1.0"and stores the exact argument vector as an array.JobResultusesschema_version: "1.0"and categorizes success, runtime failure, timeout, cancellation, and validation refusal. It includes the final normalizedProgressEvent, capture-policy output, warnings, and output existence/size facts.- profiles use
schema_version: "1.0"plus an independent positiveprofile_version. PreflightReportusesschema_version: "1.0"; human output is rendered from the same check objects.WorkflowBatchusesschema_version: "1.0"and is the shared CLI/Python envelope around a prepared plan and ordered item executions.- Batch manifests, state files, JSONL events, and ordered outcomes use independent
1.0contracts. See the batch manifest schema, example, and batch guide. - Declarative pipeline sources, state, events, preflight, and outcomes use
1.0. The pipeline schema, guide, and three golden templates define the supported contract.
Writing commands accept --result-json. Its version 1.0 envelope contains plan, preflight, ordered items, and a summary. Each item records its input/output identity, item-specific preflight, and stable JobResult. Batch image jobs therefore retain the same schema for total success, total failure, and partial success.
Run receipts use their own 1.0 JSON Schema,
redacted example, and explicit
privacy/migration contract. Human terminal output is allowed to
evolve; versioned JSON fields follow the documented compatibility and
deprecation policy.