Skills for your AI assistant
Add a skill and Claude Code, Codex, Cursor or a similar assistant learns one more job.
On-device deployment
14 skillsDeploying models to phones and embedded devices.
- metal-optimize1 installsGuides Metal backend kernel development, operator fusion, runtime scheduling and performance diagnosis on Apple GPUs.
- vulkan-optimize1 installsHelps optimize Vulkan kernels, add new Vulkan features and verify results across Adreno, Mali and Apple GPUs.
- edge-cloud-placementCompares placing a model or task on device, edge or cloud by latency, transfer, energy and recovery.
- add-new-opWalks through adding a new operator to MNN, from schema and shape calculation to backend code and unit tests.
- bugfixRoutes MNN correctness and regression bugs by symptom, covering memory, quantization, threading, GPU and cache issues.
- cpuPoints to the right CPU backend guide for ARM, x86_64 or RISC-V, whether optimizing speed or writing kernels.
- cpu-kernelGuides writing or porting CPU kernels with SIMD and assembly, registering them and checking correctness across instruction sets.
- cpu-optimizeDiagnoses why an operator or LLM run is slow on CPU and identifies which layer is the bottleneck.
- hexagon-optimizationHelps optimize, refactor and test the Hexagon DSP backend, including profiling on device and crash diagnosis.
- opencl-optimizeHelps optimize OpenCL kernels, add new OpenCL features and verify results on Android devices.
- qnn-debugDiagnoses QNN NPU errors, accuracy problems and performance issues, and guides adding new operators.
- retrospectiveAfter a nontrivial task, captures lessons learned from debugging and failed attempts for later reuse.
- support-new-llmWalks through adding a new LLM to MNN, from downloading and analyzing the model to export and alignment tests.
- test-ciRuns MNN tests and CI on host or real devices, and benchmarks LLMs on Android or iOS hardware.