Hebbian-Robotics/hflow

Migrate the Airflow HTTP transport from urllib to httpx2 about 14 hours ago

enhancement help wanted

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Migrate the EgoSuite evaluation CLI to Typer 1 day ago

enhancement help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Migrate the hflow CLI from argparse to typed Typer commands 2 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Use strict Pydantic validation for native-overlay manifest parsing 2 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Migrate the Build AI evaluation CLI to Typer 2 days ago

enhancement help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Measurement keys differing only in case silently return the wrong values in curation queries 3 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: contact_sheet writes JPEG directly to final path without atomic rename, allowing truncated crash artifacts to be cataloged as trusted evidence 3 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

checks.py still hand-rolls five numeric guards that _field_guards now covers 5 days ago

enhancement help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Good First Issue]: LeRobot importer crashes with a bare KeyError when a meta/episodes tree entry has no "path" 5 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: export-dataset-snapshot how-to still says the verifier is "not shipped here" while documenting hflow verify snapshot 5 days ago

documentation help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: _joint_motion_profile treats NaN velocity steps as compliant, causing duplicate-stamped streams to pass velocity gates and dead channels to report 0% idle 5 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: camera_video publishes scratch-bound artifact URIs as durable, causing silent byte-replacement or hard deletion on sync rewrite 5 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

LeRobot: add a reproducible real-camera import, quality, curate, and export example 5 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Repeated critical-check failures leave an episode marked ok after an intervening success 5 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]:episode cache collision via _sanitize_topic serves one camera's pixels to another, bypassing QC 6 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The suite needs an ffmpeg new enough for -fps_mode, and neither CONTRIBUTING nor any check says so 6 days ago

documentation good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The curation gate parses every query twice, and the second parse refuses nothing 6 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: catalog ui keeps empty in-memory ingest_failures after the ledger writes Parquet mid-session 6 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug] : sync-omitted stage runs measure stale scratch cache, poisoning the catalog with old evidence under new content hashes 6 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

lerobot export: frame task_index survives task-list collapse, shipping frames that reference a nonexistent task 6 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: transform: declared-but-empty camera channel is dropped, merging distinct sources and erasing provenance 6 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

video cache: concurrent remuxes share one fixed-name .tmp file, silently poisoning the frame cache with torn MP4s 6 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: Snapshot verify can read outside the handed directory via receipt paths 7 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Bug]: prepare.py: running a second factory shard silently overwrites the first corpus and stamps false provenance 7 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[BUG]: Pinned Linux FFmpeg build disappears under BtbN's retention policy 8 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

TransformConfig hand-rolls three numeric guards, and its test matches the field name rather than the message 8 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

A canonical whose chunk zstd cannot decompress is still an undiagnosed exception, not a named refusal 8 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

A preview 400 still quotes our DESCRIBE wrapper back at the caller when their SQL fails to bind 9 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

VideoImportConfig hand-rolls three numeric guards, and only one of them maps straight onto a helper 9 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

contact_sheet hand-rolls require_positive_int three times 9 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Check execution errors are silent in direct app.process() loops unless the report is inspected 9 days ago

enhancement help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

check lanes re-certify a decayed canonical episode: post-sync reads never validate chunk CRCs 9 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

camera_fps_conformance validates neither of its threshold parameters 10 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

_field_guards covers the type half only, so 49 sites hand-roll the bool guard and 7 issues have been the same bug 10 days ago

enhancement help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Integrity receipt entries are dict[str, str | int], so four use sites coerce and one of them forgets 10 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Channel.to_numpy is used in four test files and not one of its six refusals is pinned 10 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

VerificationFinding.reason is a bare str with four closed values, ten lines below a StrEnum doing the same job 11 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The fixed-GOP promise is stamped into provenance as "actually used" but is never measured on pass-through video 11 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

verify snapshot never checks integrity.content_id, so the documented deleted-member guarantee is unenforced 11 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

keyframe_interval() repeatedly rescans camera payloads, causing O(C²) keyframe parsing 11 days ago

enhancement help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

LeRobot: the data_path and video_path templates escape the info.json validation boundary 11 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Episode.channel() reads unrelated MCAP topics before channel filtering 11 days ago

enhancement help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Four sites hand-roll the same is-a-directory / not-found errno pair, two with byte-identical comments 11 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Eight camera checks repeat the same camera-selection line, and there is already a helper doing this for topics 11 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

verify snapshot accepts directories the exporter would refuse: no format identity check at the verifier boundary 11 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

doctor labels a healthy video corrupt when a decoder is unavailable 13 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

testing.py carries a byte-identical copy of _field_guards.require_int/require_float 13 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

frames_at_indices has four index guards: two are covered by an assertion that cannot tell them apart, two are not covered at all 13 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

A fractional fps in meta/info.json is floored before conversion, stretching the episode's time axis 13 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

camera_frame_stats and camera_signal_quality skip their threshold guards on a camera-less episode, and name internal fields when they do fire 13 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

[Feature]: LeRobot prepared-manifest receipts are written but nothing verifies a delivered import 13 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The HTTP 400 for SQL that is not a single SELECT is the one curation refusal nothing tests 14 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

You can pin a cut you cannot preview: preview wraps the SQL in DESCRIBE and then blames your query for the wrapper's syntax error 14 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Neither registry cap refusal is tested, and both caps are 1000 so the obvious test would be unusably slow 14 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

camera_stability accepts an invalid field of view on an episode with no cameras, because that guard lives inside the per-camera loop 14 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

fetch_uri raises the same message from two places, which reads as a dead branch and is not 14 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The camera_stability tuning knobs are undocumented, so hand-held footage reads as continuously unstable with no documented way to fix it 14 days ago

documentation good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Ingest transcodes a payload-damaged source without noticing, so the canonical episode gets a valid receipt for corrupt bytes 14 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

fetch_uri's 'names no object' refusal is the one refusal in storage.py nothing tests, and a bucket URL with no object reports a path the caller never typed 14 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

camera_stability validates one of its two new knobs: unstable_min_duration_s crashes on NaN after the whole video has been measured 14 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Snapshot integrity receipts are written but nothing reads them: no way to verify a delivered snapshot 14 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Parse LeRobot metadata into a typed internal archive model 14 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The resume path stamps a conversion receipt over reused bytes it never integrity-checked 14 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

run.json omits allow_missing_api_key, so a run against an unauthenticated endpoint cannot be replayed from its own record 14 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Three guards that cannot fire: two zero-size checks in the LeRobot resume path and an unreachable isinstance in the EgoSuite reader 15 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The Build AI example's run fingerprint excludes max_retries, so two runs with different coverage share one output directory 15 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The native overlay sample distribution is built by two copies of one helper, and they already disagree 15 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Eight LeRobot importer metadata refusals have no test, including the pagination loop guards 15 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The native overlay manifest parse stage refuses four things and none of them are tested 16 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Build AI check versions exclude max_retries and request_timeout_seconds, so evaluations that can differ share one version identity 16 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Eight of the nine refusals in OpenAICompatibleExecution are untested, including both bool-is-an-int guards 16 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

FORMAT.md documents five episode/v1 keys; the LeRobot importer writes eleven, and five of the undocumented ones gate import resume 16 days ago

documentation good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Development setup omits the C compiler required by the default test suite 16 days ago

documentation good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Dataset snapshots cannot verify table or copied-media integrity after transfer 16 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The LeRobot importer ignores the source's success label and hardcodes "true" for every episode 16 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Every refusal in the native overlay manifest validator is untested, including the schema version in the bundle digest 16 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

prepared-manifest.json records how many episodes were converted but not which ones, so a corrupt or missing episode cannot be detected from the delivery 17 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

prepared-manifest.json and the converter version it carries are never read back by a test 17 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The curated export's required 'index' column refusal is untested 18 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Ingest URIs that name the data root itself are accepted 18 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The no-B-frames constraint is refused on write but never checked by doctor or the encoder 18 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The importer cache namespacing test uses shas its own validator would reject 18 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Improve camera_frame_stats cold throughput on 1080p video without weakening per-frame evidence 18 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Validate the EgoSuite label-report envelope before reading frames 18 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The two slice-header walks still duplicate their loop and their Exp-Golomb decoder 19 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

The sync reuse gate's ffmpeg comparisons are untested: every fixture is camera-less 19 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

LeRobot video cache aliases different file indices to one MP4 19 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Slice-header scans unescape whole NALs to read two Exp-Golomb values 19 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

write_access_units_to_mp4 silently drops the reorder tail of any B-frame stream 20 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

A camera topic mixing image formats is still filed as an infrastructure failure 20 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Catalog UI shutdown installs the ui extension from the network to call stop_ui_server() 20 days ago

bug help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Model Build AI evaluation artifacts at the JSON boundary 20 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Paginate Airflow task instances before building run graphs 20 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Make EgoSuite hand-count judgment outcomes exclusive 20 days ago

enhancement good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Reject catalog evidence timestamps outside DuckDB BIGINT range 20 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Model EgoSuite evaluation artifacts at the JSON boundary 20 days ago

enhancement help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Preserve non-404 Airflow failures on stage run surfaces 20 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Centralize ingest URI parsing across CLI, server, and SDK 20 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

LeRobot cache can mix revisions when an output directory is reused 20 days ago

bug good first issue help wanted

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

A transform refusal about the recording is filed as an infrastructure failure 20 days ago

bug help wanted advanced

Hebbian-Robotics/hflow

5

Open source SDK for turning multimodal physical AI episodes into quality checked, queryable training datasets: canonical MCAP, generated Airflow pipelines, Parquet/DuckDB curation.

Python

#airflow#data-pipeline#data-quality#dataset-curation#duckdb#foxglove#mcap#parquet#physical-ai#robot-learning#robotics

Open Issues Need Help