AITER Skills Roadmap
Top 10 Claude Code skills to accelerate AMD AI operator library development. Each skill encodes team knowledge into a reusable, shareable automation.
Skills Tracker
| # |
Skill |
Priority |
Issue |
Status |
| 1 |
kernel-benchmark |
P0 |
#43 |
🔲 Not Started |
| 2 |
jit-debug |
P0 |
#44 |
🔲 Not Started |
| 3 |
quant-chain-validate |
P0 |
#48 |
🔲 Not Started |
| 4 |
op-coverage |
P1 |
#45 |
🔲 Not Started |
| 5 |
triton-vs-asm |
P1 |
#46 |
🔲 Not Started |
| 6 |
kernel-regression |
P1 |
#47 |
🔲 Not Started |
| 7 |
pr-review |
P1 |
#49 |
🔲 Not Started |
| 8 |
build-optimize |
P1 |
#50 |
🔲 Not Started |
| 9 |
api-check |
P2 |
#51 |
🔲 Not Started |
| 10 |
doc-update |
P2 |
#52 |
🔲 Not Started |
How to Contribute
- Pick a skill from the table above
- Assign yourself to the corresponding issue
- Build the skill following the spec in the issue
- Skills live in
.claude/skills/ as SKILL.md files
- Submit a PR with the skill implementation and update this tracker
Context
AITER is AMD's centralized AI operator library for ROCm/HIP GPUs. These skills cover kernel benchmarking, build optimization, quantization validation, and developer workflow automation. Target hardware: MI300X (gfx942) and MI355X (gfx950).
AITER Skills Roadmap
Top 10 Claude Code skills to accelerate AMD AI operator library development. Each skill encodes team knowledge into a reusable, shareable automation.
Skills Tracker
kernel-benchmarkjit-debugquant-chain-validateop-coveragetriton-vs-asmkernel-regressionpr-reviewbuild-optimizeapi-checkdoc-updateHow to Contribute
.claude/skills/asSKILL.mdfilesContext
AITER is AMD's centralized AI operator library for ROCm/HIP GPUs. These skills cover kernel benchmarking, build optimization, quantization validation, and developer workflow automation. Target hardware: MI300X (gfx942) and MI355X (gfx950).