v0.6.0 2026-05-13

🎨 NeMo Data Designer v0.6.0

This release comes with two major updates:

• ⚡ Async by default: The async execution engine is now the default path in Data Designer. Pipelines schedule work at the cell level, overlapping independent columns and adapting model concurrency per provider/model.

• 🔌 Plugins are stable: Plugins have moved out of experimental mode. Seed readers, column generators, and processors can now live in normal Python packages with typed configs and entry-point discovery.

What's Changed

• fix(async): pack of fixes for async engine under degraded providers by @andreatgretel in #585
• feat(config): add deterministic fingerprint for workflow configs by @nabinchha in #587
• fix(interface): don't leak YAML default provider into user-supplied list by @nabinchha in #591
• docs: Updated telemetry disclosure by @kirit93 in #595
• chore(readme): bump tokens-generated badge to 350+ billion by @johnnygreco in #593
• ci: bump semantic_pull_request from 0.93.0 to 0.94.1 by @dependabot[bot] in #596
• feat(cli): add data-designer --version by @eric-tramel in #599
• feat: make async engine the default execution path by @andreatgretel in #592
• fix: Add --org option to NGC download command by @mikeknep in #604
• ci: add graphify structural impact analysis to PR review and structure audit by @andreatgretel in #567
• fix(config): round-trip processors and profilers by @johnnygreco in #605
• feat(models): deprecate implicit default provider routing by @nabinchha in #594
• fix(engine): validate processor plugin impls by @johnnygreco in #609
• docs: graduate plugins out of experimental mode by @johnnygreco in #603
• feat(results): add export() method and --output-format CLI flag by @przemekboruta in #540
• docs: migrate documentation from MkDocs to Fern by @lbliii in #581
• fix: normalize rollout timestamps before deriving started_at/ended_at by @eric-tramel in #556
• fix: quote review-code skill argument hint by @johnnygreco in #616
• feat(cli): show version update notice by @eric-tramel in #602
• fix(docs): unbreak published Fern site by @lbliii in #615
• fix: validate subcategory parents are sampler columns by @andreatgretel in #614
• ci: bump semantic_pull_request.yml from 0.94.1 to 1.1.0 in the all-actions group by @dependabot[bot] in #621
• feat: resume interrupted dataset generation runs (sync + async engine) by @przemekboruta in #526
• fix(config): validate subcategory parent sampler type by @johnnygreco in #628
• chore(engine): rename correction-step counter for clarity by @nabinchha in #627
• fix(config): update OpenRouter vision model id by @andreatgretel in #630
• feat: let column configs declare all model aliases for the startup health check by @nabinchha in #626
• fix: harden resume checkpoint handling by @nabinchha in #624
• docs: add plan for workflow chaining by @andreatgretel in #552
• chore: update vulnerable dependencies by @johnnygreco in #631
• docs: prepare Fern docs workflow by @andreatgretel in #622
• feat(agentic-ci): decision-ready triage and daily PR fixes by @andreatgretel in #600
• feat(cli): add plugin catalog core by @johnnygreco in #618
• feat: add fair async task scheduling by @eric-tramel in #639
• docs: add "Have It Your Way" plugin dev note by @johnnygreco in #608
• fix(interface): silence registry-default deprecation when library auto-fills it by @nabinchha in #655

New Contributors

• @lbliii made their first contribution in #581

Full Changelog: v0.5.9...v0.6.0

v0.5.9 2026-04-28

What's Changed

• fix: normalize image_url blocks to OpenAI-compliant dict format by @nabinchha in #577
• docs: add VLM long-document understanding dev note and recipes by @nabinchha in #579
• docs: update resource links in VLM long-document understanding dev note by @nabinchha in #580
• ci: raise agent audit turn limit and preserve logs by @andreatgretel in #571
• feat: update default nvidia-vision model to nemotron-3-nano-omni by @nabinchha in #583

Full Changelog: v0.5.8...v0.5.9

v0.5.8 2026-04-27

What's Changed

• chore: bump pillow and python-multipart for CVEs, add SECURITY.md by @johnnygreco in #564
• fix(ci): grant permissions to reusable workflow calls by @andreatgretel in #561
• ci: bump the all-actions group across 1 directory with 5 updates by @dependabot[bot] in #558
• refactor: unify duplicate DAG construction (dag.py + ExecutionGraph) by @przemekboruta in #511
• fix: prevent sticky progress bar ghost lines from terminal wrapping by @andreatgretel in #565
• fix: Updated default reasoning model for nvidia by @kirit93 in #568
• chore: async engine readiness - blockers and polish before default by @andreatgretel in #553
• chore: remove obsolete Cerebro ignore entries by @johnnygreco in #570
• chore: add ko_KR locale to nemotron personas datasets by @johnnygreco in #572
• chore: bump lxml and nbconvert to address security advisories by @johnnygreco in #574

Full Changelog: v0.5.7...v0.5.8

v0.5.7 2026-04-17

What's Changed

• fix: restrict docs-preview workflow to actual docs changes by @andreatgretel in #515
• chore: harden CI supply chain by @andreatgretel in #517
• fix: bump pytest, aiohttp, and cryptography for security CVEs by @johnnygreco in #535
• fix: tune Dependabot config and fix DCO assistant bugs by @andreatgretel in #534
• ci: publish devnotes independently of releases by @andreatgretel in #536
• fix: async engine side-effect column propagation and collision resolution by @andreatgretel in #509
• ci: add PR hygiene automation (linked issue check + stale PR cleanup) by @andreatgretel in #521
• ci: bump the all-actions group with 5 updates by @dependabot[bot] in #539
• feat: add generic and OpenRouter attribution headers by @eric-tramel in #542
• docs: add text-to-sql dev note by @dhruvnathawani in #349
• fix: text-to-sql devnote date, images, and publish-devnotes nav by @andreatgretel in #546
• fix(ci): replace yq with Python nav patching in publish-devnotes by @andreatgretel in #548
• feat: add skip.when conditional column generation by @nabinchha in #502
• fix: use pull_request_target for agentic CI on fork PRs by @andreatgretel in #541
• docs: Added starter dev notes on push to hugging face hub by @nabinchha in #355
• fix: bridge model.generate() to agenerate() for custom columns in async engine by @andreatgretel in #545
• ci: add daily audit suites with 5 rotating recipes and scheduled workflow by @andreatgretel in #543
• feat: add RunConfig jinja rendering engine by @eric-tramel in #557

New Contributors

• @dependabot[bot] made their first contribution in #539

Full Changelog: v0.5.6...v0.5.7

v0.5.6 2026-04-09

What's Changed

• fix: use --bare and --tools in health probe CLI check by @andreatgretel in #489
• feat: add ATIF rollout ingestion by @eric-tramel in #495
• fix: replace native-model-client-hero image with corrected version by @nabinchha in #492
• docs: add skip column config option for conditional column generation (#479) by @nabinchha in #480
• chore: plan 427, PR 2 of agent-first development plan by @nabinchha in #478
• feat: add Hermes Agent rollout support by @eric-tramel in #500
• fix: prevent skill load failure when data-designer CLI is not installed by @johnnygreco in #501
• ci: add PR review workflow and recipe for agentic CI by @andreatgretel in #498
• docs: add agent rollout ingestion docs entry point by @eric-tramel in #499
• docs: add async engine dev note by @andreatgretel in #490
• fix: use non-blocking dispatch to prevent pipeline starvation by @nabinchha in #505
• feat: add Pi Coding Agent rollout seed source by @johnnygreco in #514
• fix: always return ISO-8601 from datetime postproc (#484) by @johnnygreco in #512
• fix: include multi_modal_context columns in required_columns by @nabinchha in #522
• docs: add LiteLLM supply-chain incident notice to README by @johnnygreco in #516

Full Changelog: v0.5.5...v0.5.6

v0.5.5 2026-04-02

What's Changed

• fix: Claude Code marketplace plugin structure and install docs by @johnnygreco in #458
• docs: update dev note with TL;DR tips and install instructions by @johnnygreco in #461
• chore: remove unused .claude-plugin directory by @johnnygreco in #463
• chore: async engine follow-up - rename, preview, lifecycle, progress by @andreatgretel in #456
• docs: restructure agent and contributor documentation (plan 427, PR 1) by @nabinchha in #454
• fix: address nspect vulnerability report for requests and cryptography by @johnnygreco in #475
• fix: update health checks to use new ModelFacade client API by @andreatgretel in #470
• ci: upgrade GitHub Actions for Node.js 24 compatibility by @ko3n1g in #450
• chore: reduce Greptile review noise from defensive coding suggestions by @andreatgretel in #423
• docs: consolidated seed reader documentation by @eric-tramel in #481
• fix: bump pymdown-extensions for pygments 2.20.0 compat by @eric-tramel in #482
• feat: add fr_FR locale to nemotron personas datasets by @johnnygreco in #468
• docs: add native model client dev note by @nabinchha in #465
• fix: respect max_parallel_requests in HTTP connection pool size by @przemekboruta in #460
• docs: center diagram images in native model client dev note by @nabinchha in #483
• docs: update architecture-and-performance.md to reflect AIMD changes by @nabinchha in #467
• chore: update review code skill output and tone by @nabinchha in #477
• ci: add agentic CI plan, health probe workflow, and recipe scaffold by @andreatgretel in #473
• test: add transport-wiring regression tests for #459 by @nabinchha in #485

New Contributors

• @ko3n1g made their first contribution in #450
• @przemekboruta made their first contribution in #460

Full Changelog: v0.5.4...v0.5.5

v0.5.4 2026-03-24

🔒 Note on the LiteLLM Supply Chain Incident (2026-03-24)

Earlier today, malicious versions of litellm (1.82.7 and 1.82.8) were published to PyPI containing a credential stealer that exfiltrates cloud credentials, SSH keys, and cryptocurrency wallets on any Python process startup.

Data Designer v0.5.4 removes litellm as a dependency entirely. We recommend all users upgrade.

For users on prior versions, here is our assessment:

• v0.3.0 – v0.5.3: litellm has been pinned at >=1.73.6,<1.80.12 since January — no exposure to the compromised versions.
• v0.2.2 and v0.2.3: These releases briefly carried an upper bound of <2, which in theory permitted resolution to the malicious 1.82.x versions. Both have been yanked from PyPI as a precaution.
• Realistic risk from Data Designer is very low. Exposure would require a user pinned to one of the two yanked versions and running a fresh install or dependency update during the few hours the compromised package was live this morning. That said, please verify your installed versions and upgrade Data Designer to v0.5.4 at your earliest convenience.

What's Changed

• fix: correct broken dev note links in recipe pages by @dhruvnathawani in #407
• docs: trace visualization in display_sample_record (#396) by @nabinchha in #397
• chore: simplify tutorial 4 image dataset and use default model config by @nabinchha in #403
• fix: preserve extra_body for LiteLLM to avoid UnsupportedParamsError (#409) by @nabinchha in #412
• feat: normalize validator and constraint discriminators by @johnnygreco in #414
• docs: add Open in Colab badges to tutorial notebooks by @mvansegbroeck in #391
• feat: agent CLI introspection (simplified) by @johnnygreco in #415
• fix: bump litellm lower bound to >=1.77.0 by @nabinchha in #417
• feat: Improve generation failure reporting for schema and timeout failures by @eric-tramel in #416
• feat: add built-in filesystem seed readers by @eric-tramel in #421
• feat: add AsyncTaskScheduler and RowGroupBufferManager for async engine by @andreatgretel in #404
• feat: Native OpenAI adapter with retry and AIMD throttle infrastructure by @nabinchha in #402
• refactor: simplify agent CLI to context, types, and state (#418) by @johnnygreco in #420
• feat: support 1-to-many FileSystemSeedReader hydration by @eric-tramel in #424
• fix: support nested field access in schema transform templates by @andreatgretel in #435
• chore: use uv run ruff in pre-commit hooks by @andreatgretel in #436
• test: follow up FileSystemSeedReader coverage cleanup by @eric-tramel in #432
• feat: Plan + Implementation for 392 managed storage improvements by @mikeknep in #393
• feat: Native Anthropic adapter with shared HTTP client infrastructure by @nabinchha in #426
• feat: add Data Designer skill by @johnnygreco in #434
• feat: agent rollout trace ingestion by @eric-tramel in #399
• feat: wire async task-queue scheduler into ColumnWiseDatasetBuilder by @andreatgretel in #429
• feat: Constrain HttpModelClient to single concurrency mode... by @nabinchha in #439
• docs: Updated telemetry by @kirit93 in #451
• feat: add preview review reference and update interactive iterate step by @johnnygreco in #441
• feat: add trace visualization to display_sample_record (#396) by @nabinchha in #438
• docs: agent-assisted development plan for DataDesigner by @nabinchha in #428
• feat: wire ThrottledModelClient and dual-semaphore scheduler by @nabinchha in #449
• feat: remove litellm dependency and bridge path by @nabinchha in #455
• feat: resolve data-designer command path before workflow execution by @johnnygreco in #440
• docs: Data Designer Got Skills dev note by @johnnygreco in #457

New Contributors

• @mvansegbroeck made their first contribution in #391

Full Changelog: v0.5.3...v0.5.4

v0.5.3 2026-03-12

What's Changed

• fix: cache notebook builds to avoid flaky upstream model failures by @andreatgretel in #370
• feat: canonical model client types, protocols, and LiteLLM bridge adapter by @nabinchha in #359
• fix: processor artifacts type, discovery, and loading by @andreatgretel in #366
• feat: add ExecutionGraph, CompletionTracker, and Task model for async scheduler by @andreatgretel in #356
• fix: handle discriminated unions in oneOf pruning validator by @andreatgretel in #376
• docs: account for vLLM reasoning field migration in plan 343 by @nabinchha in #377
• chore: add Claude Code skill for code review by @nabinchha in #372
• fix: replace removed DuckDB record_batch() with to_arrow_reader() by @andreatgretel in #380
• fix: patch litellm ImageURLListItem to make index field optional (#384) by @nabinchha in #385
• chore: improve test guidelines in AGENTS.md by @nabinchha in #387
• fix: raise clear error when all records are dropped during generation by @nabinchha in #383
• feat: add async generator migration with symmetric bridging and statefulness by @andreatgretel in #378
• docs: add Enterprise Text-to-SQL and Search Agent recipes by @dhruvnathawani in #395
• refactor: Decouple ModelFacade from LiteLLM via ModelClient adapter by @nabinchha in #373
• docs: search agent dev note by @dhruvnathawani in #350
• feat(cli): bootstrap default configs on CLI startup by @johnnygreco in #401
• fix: pin chardet<6 to suppress RequestsDependencyWarning by @andreatgretel in #405
• fix: add chardet<6 constraint to published engine package by @johnnygreco in #406

Full Changelog: v0.5.2...v0.5.3

v0.5.2 2026-03-04

What's Changed

• fix: repair notebook CI (dead model, missing API key, pyarrow type bug) by @andreatgretel in #348
• docs: Update top models usage chart for 1/24-2/24/2026 by @kirit93 in #353
• docs: add structured outputs SDG dev notes by @dhruvnathawani in #338
• feat: add processor plugin support by @andreatgretel in #299
• chore: plans for async generators and task-queue dataset builder by @andreatgretel in #347
• chore: plans for model facade overhaul by @nabinchha in #344
• fix: include seed dataset in builder repr for seed-only configs by @johnnygreco in #361
• chore: bump cryptography and pillow for security fixes by @johnnygreco in #364
• feat: add Streamable HTTP transport support for remote MCP providers by @nabinchha in #358
• docs: update README token badge to 150+ billion by @johnnygreco in #367
• docs: fix structure outputs blog format by @johnnygreco in #368
• chore: fix inaccuracies and improve AGENTS.md by @nabinchha in #369
• fix: include plugin column types in display_sample_record() by @3mei in #365

New Contributors

• @3mei made their first contribution in #365

Full Changelog: v0.5.1...v0.5.2

v0.5.1 2026-02-20

Data Designer now supports image generation!

What's Changed

• docs: Updated url by @kirit93 in #325
• docs: deep research trajectories with NDD and MCP tool use by @eric-tramel in #326
• refactor: callback-based processor design by @andreatgretel in #294
• feat: add image generation support with multi-modal context by @nabinchha in #317
• docs: add image generation documentation and image-to-image editing tutorial by @nabinchha in #319
• chore: move ArtifactStorage to engine/storage/ module by @nabinchha in #321
• chore: gitignore Cerebro knowledge base files by @johnnygreco in #328
• feat(engine): env-var switch for async-first models experiment by @eric-tramel in #280
• docs: Moved nav to left hand side by @kirit93 in #331
• feat: add --save-results option to preview command by @johnnygreco in #333
• chore: Improve CLI startup with lazy heavy import cleanup by @johnnygreco in #330
• feat: add allow_resize for 1:N and N:1 generation patterns by @andreatgretel in #286
• chore: address Andre's feedback on --save-results and CLI preview by @johnnygreco in #335
• chore: remove example_allow_resize.py from repo root by @andreatgretel in #337
• fix: make DropColumnsProcessorConfig idempotent and support reasoning columns by @andreatgretel in #334
• feat: add push_to_hub_from_folder classmethod for uploading saved datasets by @nabinchha in #340
• fix: handle bool, int, float in convert_to_row_element by @dhruvnathawani in #336
• feat: auto-detect ImageContext format for image-to-image generation by @nabinchha in #342

New Contributors

• @dhruvnathawani made their first contribution in #336

Full Changelog: v0.5.0...v0.5.1