A skill that teaches any coding agent to write idiomatic, correct, performant, current Mojo (Modular's language).
Mojo is young and fast-moving, and most Mojo in model training data predates 1.0. The common failure is code that looks right from memory and no longer compiles. This plugin ships an old-to-current map for Mojo 1.1 plus short, verified references, so agents target the current language rather than their training data.
- An old-to-current map applied on every Mojo task:
defnotfn,var/imm/mutnotowned/read/inout,comptimenotaliasor@parameter if, unified closures andlambda,Pointerwithunsafe_operations notUnsafePointer,Array/StringSpanrenames,max.gpunotstd.gpu,TestSuitenotmojo test,mojo precompilenotmojo package. - Rules that hold across tasks: explicit copies and moves, struct and trait model, typed
raises, explicit numeric conversion, grapheme-aware strings, interior origins. - References for memory and CPU performance (pointers, allocation, SIMD,
vectorize), GPU kernels (DeviceContext,TileTensor, warps, shared memory) and Python interop (bindings, extension modules). - A toolchain check first: when the installed
mojois older or newer than 1.1, the compiler wins.
claude plugin marketplace add agent-sh/mojo
claude plugin install mojoOr add it through the agentsys marketplace alongside the rest of the agent-sh ecosystem.
The skill activates automatically on Mojo work. Triggers include:
write a Mojo struct for a 3D vector with SIMD ops
port this Python function to Mojo
is this idiomatic Mojo?
write a GPU vector-add kernel in Mojo
why is my Mojo parallelize loop slow?
It does not activate for Python, Rust, C++, or Modular MAX serving/deployment config.
skills/mojo/SKILL.md the skill (version map, core rules, doc links)
skills/mojo/references/ memory and performance, GPU, Python interop
.claude-plugin/ Claude Code plugin + marketplace manifests
.codex-plugin/ Codex plugin manifest
agnix . # validate skill + manifests (zero errors required)
claude plugin validate . # validate marketplace and plugin manifestsMojo guidance is grounded in current upstream docs at mojolang.org/docs, not model memory - re-verify any stdlib API outside stabilized-API markers against the installed toolchain.
skill-curator- for writing and reviewing skillsagnix- the linter that validates this plugin- Part of the agent-sh ecosystem
MIT