-
Notifications
You must be signed in to change notification settings - Fork 12
Releases: flagos-ai/skills
Releases · flagos-ai/skills
Release list
FlagOS 2.1 — skills v1.1.0
Release v1.1.0
Changes since v1.0.0
Other
- Docs(kernelgen-flagos): add platform specialization routing and MCP tools (#29) by @youseeyouonedayday
- Docs(kernelgen-flagos): add MCP service prerequisite to installation section (#30) by @youseeyouonedayday
- Add auto-sync workflow for skills (#33) by @tengqm
- Feat(kernelgen-flagos): centralize MCP setup into dedicated sub-skill (#32) by @youseeyouonedayday
- Update README for line wrapping (#34) by @tengqm
- 增加KernelGen快速入门,减少介绍性文字 (#36) by @cheng874
- Docs: add kernelgen-mcp-setup sub-skill and vllm-plugin-fl-setup-flagos to READMEs (#35) by @youseeyouonedayday
- Docs(kernelgen-flagos): add kernelgen-mcp-setup.md to file descriptions (#38) by @youseeyouonedayday
- Docs(kernelgen-flagos): rename Access Token to KernelGen Token, remove BAAI references, fix VS Code typo (#39) by @youseeyouonedayday
- Fix: update MCP URL from http to https in README (#40) by @youseeyouonedayday
- Update readme en and ch (#41) by @cheng874
- Update mcp connection prompts (#43) by @cheng874
- Ci: add lint checks, auto-labeler, link checker, and secret scanner (#42) by @wbavon
- Fix: update Claude Code and Cursor plugin manifests (#44) by @wbavon
- Feat: add FlagGems PR submit and review skills (#45) by @Schopenhauer-loves-Hegel
Contributors
Thanks to all contributors who made this release possible:
- @youseeyouonedayday (7 PRs)
- @cheng874 (3 PRs)
- @tengqm (2 PRs)
- @wbavon (2 PRs)
- @Schopenhauer-loves-Hegel
Generated by Release Notes Generator
Contributors
tengqm, Schopenhauer-loves-Hegel, and 3 other contributors
Assets 2
v1.0.0
Release v1.0.0
Initial Release
Skills Included
| Category | Skill | Description |
|---|---|---|
| Deployment & Release | gpu-container-setup-flagos |
Detect GPU vendor, launch PyTorch container with correct mounts. Supports NVIDIA, Ascend, Metax, Iluvatar, AMD/ROCm. |
model-migrate-flagos |
Migrate models from upstream vLLM to vllm-plugin-FL (v0.13.0). | |
flagrelease-entrance-flagos |
Full LLM deployment pipeline orchestrator: install → verify → benchmark. | |
install-stack-flagos |
Install 5-package software stack (vLLM, FlagTree, FlagGems, FlagCX, plugin) in GPU container. | |
vllm-plugin-fl-setup-flagos |
Setup vLLM-Plugin-FL for multiple hardware backends. | |
| Benchmarking & Eval | perf-test-flagos |
Run accuracy and performance benchmarks against served models. |
model-verify-flagos |
A/B verification: isolate errors by testing with/without multi-chip stack. | |
| Kernel & Operator Development | kernelgen-flagos |
GPU kernel generation and optimization for FlagGems, vLLM, or general Triton. |
| Developer Tooling | skill-creator-flagos |
Create, validate, and scaffold new skills for FlagOS repository. |
tle-developer-flagos |
Write and optimize TLE kernels with reproducible validation. |
Other
- Add model-migrate-fl skill (#1) by @yixiaodapeng
- Add repository infrastructure and multi-IDE support (#7) by @wbavon
- Remove external repository references from READMEs (#8) by @wbavon
- Add skill-creator skill and enhance validate_skills.py (#9) by @wbavon
- Standardize skill naming with -flagos suffix (#11) by @wbavon
- Unify skill catalog with 7-category hierarchy (19 skills) (#12) by @wbavon
- Add KernelGen skills for MCP-driven GPU kernel generation (#10) by @Dongxu-H
- Update KernelGen skills for MCP-driven GPU kernel generation (#14) by @Dongxu-H
- Update KernelGen skills name adding flagos suffix (#15) by @Dongxu-H
- Add -flagos suffix validation to validate_skills.py (#16) by @wbavon
- Add GitHub Actions CI workflow (#17) by @wbavon
- Add CI workflow with Job Summary and PR comment (#19) by @wbavon
- Restructure skills catalog to 5 categories (23 skills) (#20) by @wbavon
- Fix: add all merged skills to Claude Code plugin manifest (#21) by @wbavon
- Add gpu-container-setup: a self-improving Claude Code skill for multi-vendor GPU container orchestration (#5) by @shh2000
- Add FlagRelease multi-chip LLM deployment pipeline skills (#6) by @shh2000
- Feat: add automatic Skills Catalog update workflow (#22) by @wbavon
- Add tle-developer skill (#2) by @sunnycase
- Fix: align SKILL.md name fields with directory names (#24) by @wbavon
- Docs: sync Chinese README with English version (#25) by @wbavon
- Feat(kernelgen-flagos): add optimization sub-skills (#23) by @youseeyouonedayday
- Docs(kernelgen-flagos): update READMEs with optimization sub-skills (#26) by @youseeyouonedayday
- Setup skill (#4) by @rdzhu225
- Chore: update plugin configs with all 10 skills (#27) by @wbavon
Contributors
Thanks to all contributors who made this release possible:
- @wbavon (14 PRs)
- @Dongxu-H (3 PRs)
- @shh2000 (2 PRs)
- @youseeyouonedayday (2 PRs)
- @rdzhu225
- @sunnycase
- @yixiaodapeng
Generated by Release Notes Generator
Contributors
yixiaodapeng, sunnycase, and 5 other contributors