Backplane.AgentRuntime.Command behaviour (backplane_agent_runtime v1.6.0)

Copy Markdown

Authorized command execution with environment allowlists and output caps.

Shell interpretation is a separate capability. This module never claims OS sandbox guarantees; supported cleanup is declared by the CommandPort backend.

Summary

Types

t()

@type t() :: map()

Callbacks

cancel(map, map)

@callback cancel(map(), map()) :: :ok | {:error, Backplane.AgentRuntime.Error.t()}

read(map, map, map, keyword)

@callback read(map(), map(), map(), keyword()) ::
  {:ok, map()} | {:error, Backplane.AgentRuntime.Error.t()}

start(map, map, keyword)

@callback start(map(), map(), keyword()) ::
  {:ok, map()} | {:error, Backplane.AgentRuntime.Error.t()}

Functions

cancel(command, invocation)

@spec cancel(t(), map()) :: :ok | {:error, Backplane.AgentRuntime.Error.t()}

new(namespace)

@spec new(map()) :: {:ok, t()} | {:error, Backplane.AgentRuntime.Error.t()}

read(command, invocation, job, opts \\ [])

@spec read(t(), map(), map(), keyword()) ::
  {:ok, map()} | {:error, Backplane.AgentRuntime.Error.t()}

start(command, invocation, opts \\ [])

@spec start(t(), map(), keyword()) ::
  {:ok, map()} | {:error, Backplane.AgentRuntime.Error.t()}

workspace_conflict?(arg1, arg2)