Enum ModelsCommand 

pub enum ModelsCommand {
    Package {

        source_repo: Option<String>,
        quant: Option<String>,
        target: Option<String>,
        model_id: Option<String>,
        flavor: String,
        timeout: String,
        mesh_llm_ref: String,
        experimental: bool,
        dry_run: bool,
        confirm: bool,
        follow: bool,
        status: Option<String>,
        logs: Option<String>,
        cancel: Option<String>,
        list: bool,
        update_script: bool,
        json: bool,
    
},
    Recommended {
        json: bool,
    },
    Installed {
        json: bool,
    },
    Cleanup {
        unused_since: Option<String>,
        yes: bool,
        json: bool,
    },
    Prune {
        yes: bool,
        json: bool,
    },
    Certify {
        model: String,
        report_out: Option<PathBuf>,
        json: bool,
        package_only: bool,
        api_base: Option<String>,
        prompt: String,
        max_tokens: u32,
    },
    List {
        json: bool,
    },
    Search {
        query: Vec<String>,
        gguf: bool,
        mlx: bool,
        catalog: bool,
        limit: usize,
        sort: ModelSearchSort,
        json: bool,
    },
    Show {
        model: String,
        json: bool,
    },
    Download {
        model: String,
        draft: bool,
        direct: bool,
        json: bool,
    },
    Updates {
        repo: Option<String>,
        all: bool,
        check: bool,
        json: bool,
    },
    Delete {
        model: String,
        yes: bool,
        json: bool,
    },
}

Variants

Package

Package a GGUF model for distributed inference by splitting it into layer files on Hugging Face Jobs.

Fields

source_repo: Option<String>

Source Hugging Face model ref (e.g. unsloth/Qwen3-235B-A22B-GGUF:UD-Q4_K_XL).

quant: Option<String>

Quantization variant (deprecated; prefer source refs like repo:Q4_K_M).

target: Option<String>

Target repo for the layer package (auto-derived if omitted).

model_id: Option<String>

Override model ID in the manifest.

flavor: String

HF Job hardware flavor. Use auto for the default CPU splitter baseline.

timeout: String

Requested job timeout; raised automatically by model-size minimums.

mesh_llm_ref: String

Branch or tag of mesh-llm to build in the job.

experimental: bool

Publish a public package marked experimental and open an unmerged HF catalog PR.

dry_run: bool

Explicitly keep this as a dry run. This is the default unless –confirm is set.

confirm: bool

Actually submit the HF Job. Without this, the command only prints plan, spec, and max cost.

follow: bool

Stream job logs after submission until completion.

status: Option<String>

Check status of a previously submitted job.

logs: Option<String>

Fetch logs for a previously submitted job.

cancel: Option<String>

Cancel a running job.

list: bool

List recent package jobs.

update_script: bool

Upload the latest job script to the meshllm bucket (requires org access).

json: bool

Emit JSON output.

Recommended

List recommended models from the remote meshllm/catalog.

Fields

json: bool

Emit JSON output.

Installed

List installed local models from the HF cache.

Fields

json: bool

Emit JSON output.

Cleanup

Preview or remove mesh-managed models from the Hugging Face cache.

Fields

unused_since: Option<String>

Only include models that mesh-llm has not used for the given age (for example 30d or 12h).

yes: bool

Remove the selected files instead of printing a dry run preview.

json: bool

Emit JSON output.

Prune

Remove stale derived skippy stage artifacts from the mesh cache.

Fields

yes: bool

Remove files instead of printing a dry run note.

json: bool

Emit JSON output.

Certify

Certify a Skippy layer package can be resolved, verified, and smoke-tested.

Fields

model: String

Exact layer package ref, local package dir, or catalog model ref with a package mapping.

report_out: Option<PathBuf>

Write the JSON certification report to this path.

json: bool

Emit JSON output.

package_only: bool

Stop after package resolution, integrity checks, and local stage materialization.

api_base: Option<String>

Existing mesh-llm OpenAI-compatible API base for runtime smoke gates.

prompt: String

Prompt for runtime smoke gates.

max_tokens: u32

Maximum tokens for runtime smoke gates.

List

List remote catalog models.

Fields

json: bool

Emit JSON output.

Search

Search for catalog models and downloadable GGUF/MLX artifacts on Hugging Face.

Fields

query: Vec<String>

Search terms.

gguf: bool

Filter search results to GGUF artifacts (default).

mlx: bool

Filter search results to MLX artifacts.

catalog: bool

Search only the remote meshllm/catalog.

limit: usize

Maximum number of results to show.

sort: ModelSearchSort

Sort search results.

json: bool

Emit JSON output.

Show

Show details for one exact model reference.

Fields

model: String

Exact remote catalog id or Hugging Face ref.

json: bool

Emit JSON output.

Download

Download one exact model reference.

Fields

model: String

Exact remote catalog id or Hugging Face ref.

draft: bool

Also download the recommended draft model for speculative decoding.

direct: bool

Download the exact Hugging Face file directly, bypassing catalog layer-package resolution.

json: bool

Emit JSON output.

Updates

Check or refresh cached Hugging Face repos.

Fields

repo: Option<String>

Repo id like Qwen/Qwen3-8B-GGUF.

all: bool

Operate on every cached Hugging Face repo.

check: bool

Check for newer upstream revisions without refreshing local cache.

json: bool

Emit JSON output.

Delete

Delete a specific model from local storage.

Fields

model: String

Installed model stem or Hugging Face ref (e.g. Qwen3.5-9B-BF16, org/repo, or org/repo:BF16).

yes: bool

Skip dry-run preview and delete immediately.

json: bool

Emit JSON output.

Trait Implementations

impl Debug for ModelsCommand

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl FromArgMatches for ModelsCommand

fn from_arg_matches(__clap_arg_matches: &ArgMatches) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn from_arg_matches_mut( __clap_arg_matches: &mut ArgMatches, ) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn update_from_arg_matches( &mut self, __clap_arg_matches: &ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

fn update_from_arg_matches_mut<'b>( &mut self, __clap_arg_matches: &mut ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

impl Subcommand for ModelsCommand

fn augment_subcommands<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate Self via FromArgMatches::from_arg_matches_mut

fn augment_subcommands_for_update<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate self via FromArgMatches::update_from_arg_matches_mut

fn has_subcommand(__clap_name: &str) -> bool

Test whether Self can parse a specific subcommand