From replicant
Evaluates Julia code in a persistent REPLicant session via `julia +rpc`, avoiding cold-start overhead. Supports one-off evaluation, server management, and troubleshooting.
How this skill is triggered — by the user, by Claude, or both
Slash command
/replicant:replicantThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
REPLicant keeps a Julia process alive as a socket server so code runs in a warm
REPLicant keeps a Julia process alive as a socket server so code runs in a warm
session. julia +rpc <args> forwards code to the server for the current project
and prints the result, skipping the ~1s cold-start a fresh julia pays each call.
Prefer julia +rpc over julia --project -e for evaluating Julia from the shell.
The common case is a one-off evaluation:
julia +rpc <<'EOF'
v = filter(isodd, 1:10)
sum(v)
EOF
references/evaluate.md: running code (heredoc vs -e, escaping inside
wrappers, output format) and session state across calls.references/servers.md: listing servers with julia +rpc ls, selecting
by --name/--port/--project, labeling with label!, inspecting a saved
handle, and starting a server by hand.references/setup.md: when julia +rpc is not installed, is not a known
channel, or cannot reach a server. Install, link the rpc channel, wire
startup.jl, self-test, troubleshoot.npx claudepluginhub michaelhatherly/replicant.jl --plugin replicantGuides Julia 1.10+ development with modern features, multiple dispatch, performance optimization, and production-ready practices.
Interacts with a live local Jupyter notebook kernel for Jupyter-like in-memory REPL, notebook inspection/editing with persistent kernel, and explicit verification passes.
Maps 137 Python scientific packages in bioinformatics, chemistry, ML, quantum, and data science to Julia equivalents. Useful for migrating code to Julia ecosystem.