Skip to content

Releases

v0.1.1

Choose a tag to compare

@github-actions github-actions released this 01 Nov 18:59

Workbench v0.1.1

A local-first terminal workbench for files: 65 operations across images, audio,
video, documents, and general files. One binary, optional engines, and four ways
to drive it — the TUI, the CLI, a local HTTP API, and an MCP server.

Install

brew install solomonadzape95/tap/workbench

Or use the installer:

curl -fsSL https://raw.githubusercontent.com/solomonadzape95/workbench/main/install.sh | sh

Release archives for macOS and Linux (arm64 and amd64) and checksums.txt are
below. Go users can run
go install github.com/solomonadzape95/workbench/cmd/workbench@latest.

Highlights

  • 65 operations across image.*, audio.*, video.*, doc.*, and
    file.*, including background removal, local Whisper transcription, PDF OCR,
    PDF compression/merge/split, image-to-PDF, watermarking, deduplication, and
    checksum verification.
  • Pipelines — build a chain in the TUI, save it with ctrl+s, and re-run it
    from the palette or with wb run --pipeline.
  • Scriptable CLI — wb run, wb watch, wb ops, wb engines,
    wb pipeline, wb history, wb doctor, and shell completions. Machine-readable
    output with --json, and a non-zero exit when a job fails.
  • Agent-ready — wb serve exposes the same runner over HTTP with SSE
    progress and artifact downloads; wb mcp exposes every operation as a typed
    MCP tool over stdio.
  • Local by default — no uploads, accounts, or second API key. ffmpeg,
    yt-dlp, whisper.cpp, LibreOffice, pandoc, Tesseract, Ghostscript, pdf2docx,
    poppler, rembg, Real-ESRGAN, and OpenCode are detected only when needed.
  • TUI — palette, path/alias completion, parallel jobs, persistent history,
    logs, output review/undo, previews, a built-in editor, eight themes, settings,
    and contextual help.

Notes

  • macOS and Linux binaries. Windows is not supported yet.
  • Optional engines enable their operations; wb doctor shows what is ready and
    how to install the rest.
  • doc.detect_ai is diagnostic, not proof. Read
    docs/ai-detector.md
    before treating a score as evidence.
  • The Homebrew formula is maintained in
    solomonadzape95/homebrew-tap.

Thanks

If Workbench saves you a batch job, star the repo and tell me which operation
you want next.

v0.1.0

Choose a tag to compare

@github-actions github-actions released this 31 Oct 11:10

Changelog

  • 4e4d046 chore(ci): keep the GitHub Actions workflow as a deployable template
  • 2352b5c chore(lint): satisfy the Go 1.26 linter
  • c63e88b chore: add version metadata package
  • bdd1586 chore: bootstrap Workbench project
  • 40d72a3 ci: activate the GitHub Actions workflow
  • 9cf7bda docs(lint): add package and exported API comments
  • ac15b2c feat(api): HTTP API, MCP server, OpenAPI schema, and Go client
  • d1fb86c feat(cli): compose the application and add MCP-aware commands
  • 7790aec feat(core): operation contracts, engines, config, and runner
  • 91ad8ac feat(ops): file, media, document, and AI-detection operations
  • a994a21 feat(tui): screens, review, statistics, presets, and golden frames
  • e0e89a3 feat: add the Workbench website, dither logo, and release packaging
  • ea03abe fix(ci): make golden frames hermetic and use a Go 1.26 linter