Struct Execute 

pub struct Execute {
    pub bypass_local_limit: bool,
    pub shell: ExecuteShell,
    pub script: Option<String>,
    pub script_file: Option<String>,
    pub script_object: Option<String>,
    pub timeout: String,
    pub run_as: RunAs,
    pub cwd: Option<String>,
}

Fields

bypass_local_limit: bool

Critical jobs can bypass the PC limit and consume no slot.

shell: ExecuteShell

script: Option<String>

Inline script body. Mutually exclusive with script_file and script_object; exactly one of the three must be set (enforced by Execute::validate_script_source at the write-side parse boundaries — kanade job create and POST /api/jobs).

Empty string is treated as unset so operators can swap to a script_file: / script_object: alternative just by commenting out the body, without having to also drop the script: key entirely.

script_file: Option<String>

Repo-local file path resolved by the operator-side CLI at kanade job create time. The CLI reads the file, slots its contents into script, and clears this field before POSTing — so the backend / agents never see script_file in stored manifests. SPEC §2.4.1.

The resolver shipped with #210: kanade job create / kanade job validate inline this field end-to-end. Because resolution is CLI-side (it needs the operator’s filesystem), POST /api/jobs rejects a manifest that still carries it (#918) — a stored script_file job would 400 at every exec. Inline the script or use script_object when writing through the API / SPA editor.

script_object: Option<String>

Object Store reference (<name>/<version>) into the scripts bucket (OBJECT_SCRIPTS). Agents fetch the body at Execute time via /api/script-objects/{name}/{version} and cache it locally. SPEC §2.4.1.

Fully wired (#210/#211): the backend resolves the digest at exec submission (api::exec::resolve_script_source), the agent fetches + sha-verifies + caches the body (script_cache), and kanade script CRUDs the store. Unlike script_file: (inlined CLI-side, git-managed), this keeps the body in versioned, digest-pinned object storage — the ops-managed counterpart.

timeout: String

humantime duration string (e.g. “30s”, “10m”). Script-intrinsic — represents how long this script reasonably takes to run.

run_as: RunAs

Token + session combination the agent uses to launch the script (v0.21). Default = RunAs::System (Session 0, LocalSystem privileges, no GUI; root on macOS) — matches pre-v0.21 behavior. user / system_gui run in the logged-in console user’s session on Windows and macOS, and fail when nobody is logged in; Linux agents skip them.

cwd: Option<String>

Working directory for the spawned child (v0.21.1). When unset, the child inherits the agent’s cwd — on Windows that means %SystemRoot%\System32 for the prod service, which is almost never what operators actually want (a macOS run_as: user job starts in the user’s home instead: the daemon’s own cwd is its 0700 data dir). Use an absolute path; relative paths are passed through to the OS verbatim. The agent expands a leading ~ to the home of the identity the job runs as (the user for run_as: user), and on Windows %FOO% from that identity’s environment — so %PROGRAMDATA% works for run_as: system, ~\src / %USERPROFILE% for run_as: user. A macOS agent expands only ~.

Implementations

impl Execute

pub fn validate_script_source(&self) -> Result<(), String>

Enforce that exactly one of script / script_file / script_object is set. Called at the write-side parse boundaries (CLI kanade job create + backend POST /api/jobs) so ambiguous YAML is rejected before it reaches the JOBS KV. Read paths (projector, agent scheduler, list endpoints) skip this check — they only ever see what the write path already validated.

Trait Implementations

impl Clone for Execute

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Execute

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Execute

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl JsonSchema for Execute

fn schema_name() -> Cow<'static, str>

The name of the generated JSON Schema.

fn schema_id() -> Cow<'static, str>

Returns a string that uniquely identifies the schema produced by this type.

fn json_schema(generator: &mut SchemaGenerator) -> Schema

Generates a JSON Schema for this type.

fn inline_schema() -> bool

Whether JSON Schemas generated for this type should be included directly in parent schemas, rather than being re-used where possible using the $ref keyword.

impl Serialize for Execute

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.