K4sh | Plugins

Yes, k4sh has a native backtester. But the paper-trader plugin simulates live market conditions including slippage, partial fills, and even simulated exchange API downtime. It is invaluable for stress-testing your strategy before going live.

k4sh plugins are modular, hot-loadable extensions that add new commands, automation routines, and third-party tool integrations to the core k4sh environment.
Designed for red teamers, pen-testers, and automation engineers, the plugin system allows you to:

Standard k4sh does not include machine learning. The k4sh-ml-inference plugin bridges this gap. It allows you to load pre-trained ONNX or TensorFlow Lite models directly into your trading loop.

Why traders love it:

# A k4sh plugin is just an executable (any language)
# that implements this contract:

k4sh_plugin_meta() 
echo '
"name": "git-radar",
"version": "0.2.0",
"hooks": ["prompt", "precmd"],
"requirements": ["git>=2.0"]
'



k4sh_plugin_run() 
local hook="$1"
case "$hook" in
prompt)   echo "$(git branch --show-current 2>/dev/null)" ;;
precmd)   git fetch --quiet --dry-run 2>/dev/null ;;
esac

Turn k4sh into your ultimate modular command center. Seamlessly integrate custom scripts, expand functionality, and streamline your workflow.

K4SH plugins are modular scripts or compiled binaries that extend the core functionality of the K4SH shell environment. Designed for speed and minimalism, K4SH allows dynamic loading of plugins to add features like:

Unlike monolithic shells, K4SH loads plugins on-demand to keep memory footprint low. k4sh plugins

Some plugins depend on others. For example, k4sh-visual-dashboard requires k4sh-nexus to be active to show aggregated data. The plugin manager resolves these automatically, but always read the plugin’s documentation.

  • Keep init.sh idempotent: sourcing multiple times should not duplicate definitions.
  • Namespace functions and variables to avoid collisions (prefix with plugin name).
  • Provide minimal startup footprint; defer heavy work.
  • Offer completions placed under completions/ using the shell’s completion API.
  • Declare dependencies in plugin.toml and fail gracefully if missing.
  • Include tests where possible (unit scripts or basic smoke tests).

    • Example minimal plugin (plugin.toml + init.sh)

  • init.sh:

    • Example lazy-command stub pattern (in k4sh core) Yes, k4sh has a native backtester

  • plugin’s init.sh then defines real hello() replacing the stub.