llmff v0.1.4

Roadmap-completion release for agent and automation workflows. This release
keeps the GitHub Release asset lane from v0.1.3 and finishes the seven
forward-looking tracks that make llmff a dependable FFmpeg-style runner for
LLM inference pipelines: stable process semantics, explicit inputs,
reproducible manifests, inspectable execution, local supervision, integration
readiness, and predictable distribution.

Supported Install

cargo install --git https://github.com/syndicalt/llmff --tag v0.1.4 llmff

Included Since v0.1.3

• Stable subprocess contracts for successful runs, validation failures, stage
  failures, provider/runtime failures, local data failures, interrupted runs,
  batch failures, stdout ownership, event streams, traces, checkpoints, and
  additive failure_kind values.
• Manifest reproducibility guidance and gates for inspect JSON output, resolved
  inputs and outputs, stage order, backend aliases, model ids, plugin
  dependencies, cache policy, checkpoint/resume policy, and the parked lockfile
  decision.
• Local observability guidance, same-run supervision example, OpenTelemetry
  bridge design, and local gates that keep telemetry file-based and opt-in.
• Agent workflow documentation with runnable Python subprocess supervisor,
  batch supervisor, and Node.js streaming supervisor examples.
• The agent runner adoption guide covers preflight, dispatch, supervision,
  artifact collection, retry decisions, and metadata payload boundaries.
• Provider smoke readiness, provider capability docs, plugin protocol fixtures,
  plugin registry guidance, plugin trust review, and static conformance checks.
• Distribution trust docs, release trust manifest generation, signed apt
  repository design, unsigned Windows/macOS installer status, and package
  metadata kept support-ready but unpublished.
• Ecosystem readiness gates for manifest contracts, trace/event streams, CLI
  JSON output, plugin protocol, provider onboarding, agent embedding, agent
  runner adoption, package-manager metadata, release assets, and the
  OpenTelemetry bridge.
• Release preflight now validates ecosystem readiness before release tagging;
  release preflight is the local gate before creating or pushing the tag.
• Governance, stability, release compatibility, and deprecation policies for
  the documented contracts.

Packaged Artifacts

Release-tag CI is expected to publish:

• Linux x86_64 .tar.gz archive, .deb package, and Arch PKGBUILD plus
  llmff-0.1.4-arch.SRCINFO metadata.
• macOS Apple Silicon and Intel .tar.gz archives and unsigned .pkg
  installers.
• Windows x86_64 unsigned .zip archive and unsigned MSI installer.
• Adjacent SHA-256 checksums for generated binary artifacts.

Manual workflow dispatch keeps generated files as Actions artifacts instead of
GitHub Release assets.

Known Limitations

• This is a pipeline runner, not a native inference kernel, model converter,
  serving platform, or full agent framework.
• Windows and macOS native artifacts are unsigned in this release.
• Package-manager distribution through Homebrew, winget, Scoop, apt
  repositories, or an official AUR package remains parked until maintainers
  decide those channels are support-ready.
• Authenticode signing, Apple Developer ID signing, and notarization remain
  deferred paid distribution tracks.
• Live provider smoke jobs remain opt-in and require maintainers to configure
  secrets and runner expectations explicitly.
• Homebrew, winget, Scoop, and AUR prototype metadata remains pinned to the
  latest verified package-manager asset set until v0.1.4 artifacts are
  published and maintainers decide those channels are support-ready.

Verification

Release verification should pass before tagging:

scripts/release-preflight.sh v0.1.4
scripts/smoke-install.sh --git https://github.com/syndicalt/llmff --tag v0.1.4