For AI agents: the complete documentation index is available at https://a3s-lab.github.io/ash/v0.1.0/en/llms.txt, the full documentation bundle is available at https://a3s-lab.github.io/ash/v0.1.0/en/llms-full.txt, and this page is available as Markdown at https://a3s-lab.github.io/ash/v0.1.0/en/guide/index.md.
  • English
  • v0.1.0
  • Get started with ash

    ash is an AI Native Shell for Coding Agents. It accepts typed ASH/1 requests, runs work across bounded Tokio I/O and Rayon CPU planes, and returns canonical ASON instead of terminal text formatted for people.

    Current status

    This is the v0.1.0 documentation checkpoint for source commit d875661. The source, installers, and six-target workflow are implemented and pass Linux, macOS, and Windows CI, but this is not a supported binary Release. Release credentials are not provisioned and no supported signed binary is published.

    Choose an integration

    ScenarioEntry pointContract
    Long-lived Agent Harnessash rpcFrames, handshake, concurrent requests, cancellation, and result references
    One typed callash runReads one bare ASON request from standard input
    ASON canonicalizationash asonValidates and emits the unique canonical representation
    Install and upgradeinstall.sh / install.ps1Target detection, verification, transactional activation, and uninstall
    Development validationCargo source buildBuilds the a3s-ash package with the checkpoint lockfile

    Minimal one-shot request

    Send a canonical ASON request to ash run:

    i:req-1
    o:read
    b{r,o}:1024,256
    a{p,s,m}:"README.md",0,512
    ash run < request.ason

    The response is canonical ASON as well. Errors contain stable diagnostic codes and typed fields—not help text, color, progress bars, or human suggestions.

    1. Installation for platforms, targets, and the current release boundary.
    2. Architecture for the dual execution planes and governor.
    3. ASH/1 for handshake, framing, requests, and responses.
    4. ASON for canonical encoding, columns, and references.
    5. Security for capabilities, permits, transactions, and signed updates.