Module command_graph

Expand description

The canonical command graph: every command this repository offers, from whichever program offers it, with where each one is projected.

  clap tree            share/commands.yaml         just --dump
  (the executable)     (the shell tool)            (the workflows)
        │                     │                          │
        └──────────┬──────────┴──────────┬───────────────┘
                   ▼                     ▼
             contributors          capability registry
                   └──────────┬──────────┘
                              ▼
                       CommandGraph  ── fingerprint
                              │
   ┌──────────┬───────────────┼───────────────┬──────────────┐
   ▼          ▼               ▼               ▼              ▼
 bridge   completion      MCP · HTTP       Cockpit         docs

§What this is not

It is not a registry. Nothing is declared here and nothing registers with it: every node is read from a declaration that already existed and is already authority for what it states — clap for the shape of the native command line, the shipped command registry for the shell tool, the workflow runner’s own dump for the recipes. Adding a command to any of those puts it in the graph, and every projection follows, because a projection asks the graph rather than keeping a list.

§Dependency direction

The graph depends on the contributors and on the repository environment. Nothing it projects depends on it in return: the generated bridge, the completion engine, the capability module that serves it and the reference pages are all consumers. A projection that fed the graph would make the graph a rendering of its own output.

Re-exports§

pub use model::ArgumentSpec;
pub use model::Availability;
pub use model::CommandGraph;
pub use model::CommandId;
pub use model::CommandNode;
pub use model::Diagnostic;
pub use model::Effect;
pub use model::Execution;
pub use model::Interactivity;
pub use model::Origin;
pub use model::Projections;
pub use model::Provenance;
pub use model::Requirement;
pub use model::Secrecy;
pub use model::Severity;
pub use model::Surface;
pub use model::ValueChoice;
pub use model::ValueSource;
pub use model::SCHEMA;

Modules§

bridge: The workflow bridge: the command graph projected into recipes.
complete: One completion engine, for every surface.
load: Loading the graph, at two speeds.
model: The typed model of the command graph: what a command is, and what every projection of it may say.
native: The native command line as command-graph nodes.
policy: The exposure policy: which surfaces carry a command, derived from what the command is.
semantics: The semantics of the native command line that its declaration cannot carry.
shell: The shell adapters: the only shell code in the completion, and it knows nothing.
tool: The shell tool’s commands as command-graph nodes.
workflow: The repository’s own workflows as command-graph nodes.

Structs§

Inputs: What a build reads.

Functions§

build: Build the graph.
entrypoints: The commands a newcomer is offered, in a deterministic order.