CLI command correction: six recorded examples
The downloads contain six original inputs, their expected labels and one recorded API observation per input revision. Disagreements stay visible.
Provide the invalid token, intended operation and command help. Suggest a unique listed match, or leave it unchanged when aliases, paths or meaning are unresolved.
Semantic correction uses the help descriptions, not just similar spelling. An invalid word such as inventory can refer to a listing command even when the two strings look different.
The correction boundary matters. An alias, plugin or file path can be valid outside a built-in subcommand list, and an ambiguous word must not become a destructive action by guesswork.
The linked jevyoumean project explores semantic CLI suggestions. This independent example uses a fixed fictional help list and only displays a label; it does not invoke a shell or pass arguments to a command.
This is an independently authored text demonstration inspired by the linked X discussion. Its rules and synthetic inputs are ours. It does not reproduce the author’s application, verify reported outcomes, or perform an external action.

Use the sample to understand the rule, then test your own input for cli command correction.

Provide the invalid token, intended operation and command help.

Open the decision rules and keep the intended distinction between list, remove, restore, No correction. Edit the choices only when testing your own rule.

Use a sample first, then try your own input. Compare the selected label with the other returned probabilities.

Review uncertain or unexpected results before using a result. The label is a decision about the supplied information, not an executed action.
| Original input | Observed label | Leading probability |
|---|---|---|
| Inventory intent | list | 100% |
| Undo intent | restore | 100% |
| Explicit remove intent | remove | 99.0% |
| Possible alias | No correction | 73.0% |
| Path-like input | No correction | 89.0% |
| Unclear destructive word | No correction | 99.0% |
These are six original synthetic inputs, tested once per request revision. The percentage is the leading option probability, not measured accuracy. Authored expectations and any disagreements remain in the downloadable evidence. 6 of 6 results matched the authored fixture expectations.Verified
The downloads contain six original inputs, their expected labels and one recorded API observation per input revision. Disagreements stay visible.
The single-request SDK runner uses your private TYPESAFE_API_KEY. The offline consumer reads saved observations, checks request hashes and prints a pending output. It does not call the model or change any external system.
# Offline: inspect the downloaded saved observations. No API key is needed.
python preview_choice_workflow.py samples.json saved-api-results.json
# Optional live request: install the SDK and set TYPESAFE_API_KEY privately.
npm install @typesafe-ai/sdk@0.6.0
node run-choice-sample.mjs samples.json case-1
# The live response prints to the terminal; it does not update saved-api-results.json.The input contract and these limits define what the result means.
No. All outputs are labels in a text demonstration. No command is executed and no resource is selected.
The supplied help list may not describe user extensions. Without that context, rewriting the token could change a valid user command.
It describes the distribution over the supplied choices for this request. These six synthetic examples are functional tests with authored expectations, not a production accuracy benchmark.
No. The demo displays a label and distribution. The downloadable consumer produces a pending preview packet. Application code and a separate authorization policy would control any real action.
Edit the example above to see how a changed fact affects the decision, or open the Playground for another task.