bijux-dag-cli 0.4.0

Installable command-line package for validating, running, replaying, and inspecting Bijux DAG graphs.
bijux-dag-cli-0.4.0 is not a library.

bijux-dag-cli

Crates.io Rust docs License: Apache-2.0 CI Status GitHub Repository

Repository docs bijux-dag-cli docs

bijux-dag-cli installs the bijux-dag executable. It is the public command package for the DAG product.

bijux-dag v0.4.0 is a local-first DAG runtime for reproducible workflows with explicit graph contracts, deterministic execution records, verified artifacts, cache explanation, and replayable run bundles. This crate is the installed command surface for that promise.

Install it when you want the standalone DAG command surface:

cargo install bijux-dag-cli
bijux-dag --help

For most users, this is the package to start with. It gives you the supported local DAG workflow surface without asking you to assemble the lower-level DAG crates yourself.

Release Status

  • public crate on the v0.4.0 DAG release line
  • installs the stable operator-facing bijux-dag binary
  • does not promote experimental, simulated, or internal namespaces into the default public contract

Stable Operator Boundary

The supported release boundary is the visible bijux-dag --help surface: validate, plan, run, replay, runs, artifact, artifact-inspect, diff, explain, verify, doctor, cache, version, commands, and completions.

Experimental routes remain available by explicit path for repository-owned workflows, and bijux-dag commands --lane experimental is the deliberate inventory surface for that lane. Simulated and maintainer namespaces require explicit lane inventory through bijux-dag commands --lane simulated or bijux-dag commands --lane internal, plus execution opt-in through BIJUX_DAG_ENABLE_SIMULATED=1 or BIJUX_DAG_ENABLE_INTERNAL=1.

Container-backed nodes are part of the stable local execution surface. When a supported engine such as Docker is available on PATH, bijux-dag run executes the container step, retains stdout and stderr, and records engine and image identity in the retained trace. When the engine is missing, the run fails as an infrastructure error rather than silently degrading to another execution mode.

Branch-backed workflows are also part of the stable local execution surface. When a DAG uses a branch node, retained runs record the selected decision, the skipped lane, and the join-node trigger evaluation so operators can inspect the execution path directly from run evidence.

What It Provides

  • the bijux-dag binary entrypoint
  • thin startup wiring, process initialization, and exit mapping
  • delegation into bijux-dag-app for actual command behavior
  • shell completion generation for the installed executable
  • lane-scoped command discovery for stable, experimental, simulated, and maintainer-only route inventories

What It Does Not Own

  • graph semantics
  • runtime execution logic
  • artifact persistence rules

If the question is about route behavior rather than process startup, the next place to read is usually bijux-dag-app.

Representative Workflows

  • Executable Examples maps the repository-backed hello, file-processing, cache, replay, failure, branch, and container proofs to their expected outputs.
  • File Processing Workflow demonstrates a host-shell artifact workflow with replay and promotion.
  • Cache Behavior Workflow demonstrates stable cache verification, explicit-path cache-miss explanation, selective invalidation, and corruption refusal.
  • Reproducibility Model explains the retained identity surfaces behind cache verification, replay, export bundles, and artifact comparison.
  • Data Pipeline Workflow demonstrates retained-run comparison and changed-input attribution.
  • Branching Bulletin Workflow demonstrates retained branch decisions, skipped lanes, join-trigger evidence, and replay stability.
  • Compliance-Gated Bulletin Workflow demonstrates transient retry evidence, focused replay repair, and strict verification after recovery.
  • Container Packaging Workflow demonstrates mounted container inputs, retained outputs, and recorded image identity.

Internal schedule and backfill guides remain available in the DAG handbook for deliberate maintainer work with BIJUX_DAG_ENABLE_INTERNAL=1, but they are not front-door examples for the public bijux-dag package.

Internal Documentation

Related links