已通过算法建模与仿真验证v3.0.0

边缘 AI FPGA 加速助手

规划 FPGA AI 加速器的算子划分、量化、缓存复用、DDR 访问、AXI 控制和精度/吞吐验证。适合工业检测、视频分析、低功耗边缘推理和算法原型验证，重点解决“AI 加速项目常在量化精度、片上缓存、DDR 带宽、批处理和软件接口之间反复返工”这类真实 FPGA 项目问题。输出 AI 加速器映射表、缓存与带宽预算和可执行的后续动作。

查看 Skill 详情

Benchmark

96.7%

通过率

检查项

low

风险等级

publishable

复核结论

自动检查

SKILL.md format and section validation

通过

skills/edge-ai-fpga-accelerator/SKILL.md

Required frontmatter and sections are present.

Hardcoded secret scan

通过

skills/edge-ai-fpga-accelerator

No private key, cloud key, token, or long generic secret matched.

High-risk behavior scan

通过

skills/edge-ai-fpga-accelerator

No recursive deletion, cloud metadata access, encoded shell, or unreviewed transfer matched.

Declared dependency inventory

通过

skills/edge-ai-fpga-accelerator

No runtime dependency manifest is included in this Skill package.

Sandbox dry-run readiness

通过

skills/edge-ai-fpga-accelerator

Package is documentation/reference only, so runtime sandbox is marked as dry-run ready.

Benchmark evidence completeness

通过

skills/edge-ai-fpga-accelerator/SKILL.md

Score 95, level A+, pass rate 96.7%.

Human review gate

通过

skills/edge-ai-fpga-accelerator/SKILL.md

Status is reviewed.

Benchmark 套件

Format and metadata fixtures12/12

content/audit/evidence/edge-ai-fpga-accelerator/bm-fmt.json

IC workflow scenario cases30/31

content/audit/evidence/edge-ai-fpga-accelerator/bm-scenario.json

Safety and guardrail cases12/12

content/audit/evidence/edge-ai-fpga-accelerator/bm-safety.json

Regression and replay cases5/6

content/audit/evidence/edge-ai-fpga-accelerator/bm-regression.json

包盘点

包哈希: sha256:9afd9ba990f11fea
文件数: 7
可执行文件: 0
复核结论: publishable
复核团队: IC Hub 审核团队
复核时间: 2026-06-11

已知限制与下一步

适用边界

审核结论只覆盖 Skill 包内容、安装计划和公开样例，不替代真实 FPGA 项目的上板测试、客户验收和安全审批。

Ready for controlled download once account authorization is connected.