pyclisteno

A learned shortcut, hint, and completion layer for Python CLIs built on Click or Typer.

Stenography compresses language by rule rather than by lookup table, which is why trained stenographers reach speed without memorising a second vocabulary. clisteno does the same thing to a command tree: the short form of a command is a literal prefix of the long form, so typing dectl sa gl so ru is typing dectl salesdata glue source-copy run with less of it, not typing an alias that has to be learned separately.

What it provides

Three surfaces over one model, each independently opt-in:

• Teaching — the short form and its description shown beside the long form in help output, so ordinary use trains the fast path.
• Prediction — grammar-aware ghost text in the shell, driven by a cached index rather than by history.
• Compression — a resolver that expands a prefix sequence into the full command.

Attaching it

Enrollment is one call after the command tree is complete. It is not a decorator on each command, because a tree assembled at runtime from config has no functions to decorate:

from pyclisteno import attach

app = typer.Typer()
# ... build the tree, including anything driven by config ...
attach(app)

Removing the library is deleting those two lines. Nothing about the CLI changes: with attach() as written above, stdout, stderr and exit codes are byte-identical without it, at every level of the tree and for every --help, which the test suite asserts rather than assumes.

Teaching is the one surface that changes what a CLI prints, so it is asked for explicitly:

attach(app, teaching=True)

Each row in a parent's help then carries the short form of the command it names.

run      (ru) Run the thing, and keep running it until something says otherwise.
runs     (runs) List previous runs.
destroy  Delete everything, permanently and without confirmation.

runs shows its whole name because run is a strict prefix of it, and destroy shows no short form at all — see @no_shortcut below.

Decorators exist only for exceptions:

@app.command()
@shortcut('ru')  # pin, rather than accept the computed prefix
def run(): ...


@app.command()
@no_shortcut  # keep a destructive verb off the fast path
def destroy(): ...

How prefixes are assigned

Each node gets the shortest prefix of its own name that is unambiguous among its siblings, so run and review become ru and re. Assignments are recorded and grandfathered: once a prefix has been handed out it is never shortened, lengthened, or reassigned, because the whole point is protecting a sequence already in someone's fingers. Resolution is longest-match, which is what lets an incumbent r coexist with a later re. A removed command's prefix is retired permanently rather than recycled — a recycled sequence still works and silently does something else, which is the worst failure available here.

Where it keeps things

The config file sits beside the tool's own config and is never merged into it, so the tool never parses shortcuts and never needs to know this library exists. Its keys are node paths and its values are that node's prefix at its own level, which is also how the ledger is keyed:

[shortcuts]
run = "ru"
"glue nightly" = "ni"

A pin that is unusable — not a prefix of the command's own name, or already spoken for — is dropped and the prefix computed as if it were absent. Nothing in that file can stop the CLI starting, including a syntax error in it.

Related

goclisteno and bashclisteno implement the same grammar schema and the same assignment algorithm for their ecosystems, in the way goselfupdate / pyselfupdate / bashselfupdate share one release contract.