genblaze-cli 0.3.5

CLI for genblaze manifest operations

genblaze-cli

Command-line toolkit for inspecting, verifying, and indexing genblaze AI-generated-media provenance manifests.

genblaze-cli is the companion CLI for genblaze — the Python SDK for generative AI pipelines across video, image, and audio. It lets anyone (not just Python developers) audit AI-generated media: extract the embedded provenance manifest from an MP4 / PNG / MP3, verify its manifest hash and output sha256 coverage, replay a run, or index manifests into Parquet for downstream analytics.

Why genblaze-cli

• Audit AI-generated media in one command — genblaze verify video.mp4 confirms a file's manifest hash hasn't been tampered with and all output assets declare sha256.
• Works on any genblaze output — PNG, JPEG, WebP, MP4, MP3, WAV with embedded manifests.
• Analytics-ready — genblaze index emits partitioned Parquet tables (runs, steps, assets) for BI tools and data warehouses.
• Zero provider dependencies — reads manifests; doesn't call any AI API.
• Shell-friendly — non-zero exit codes on verification failure, pipeable JSON output.

Install

pip install genblaze-cli

Installs the genblaze console script.

Usage

genblaze --help

genblaze extract video.mp4                # Extract embedded manifest → stdout (JSON)
genblaze extract video.mp4 -o m.json      # …or to a file

genblaze verify video.mp4                 # Verify manifest hash + output sha256 declarations
genblaze verify video.mp4 --fetch         # …and re-hash the actual asset bytes
genblaze verify manifest.json             # Or verify a standalone manifest file

genblaze replay manifest.json             # Show what a replay would do (dry run)

genblaze index manifest.json -o data/     # Index into partitioned Parquet tables

Exit codes are non-zero on verification failure; safe to drop into CI pipelines, release checks, or content-moderation workflows. By default verify checks the manifest only; --fetch also downloads each output asset (through the same SSRF-hardened transfer path the pipeline uses) and compares its bytes against the declared sha256. Local pipelines that wrote assets outside the system temp directories (e.g. output_dir=./out) can admit them with --allowed-root <dir>.

Typical flow

# 1. Someone ships you an AI-generated video
genblaze extract delivered-ad.mp4 -o manifest.json

# 2. Confirm it hasn't been tampered with
genblaze verify delivered-ad.mp4

# 3. Index into your analytics warehouse
genblaze index manifest.json -o s3://analytics/genblaze/

Documentation

• Main repo: https://github.com/backblaze-labs/genblaze
• CLI reference: https://github.com/backblaze-labs/genblaze/tree/main/cli

Related packages

• genblaze-core — the pipeline SDK that produces these manifests
• genblaze-s3 — durable storage on Backblaze B2, AWS S3, Cloudflare R2, MinIO

License

MIT