agentmailer-security
Design or review secure AgentMailer integrations, authorization boundaries, prompt-injection defenses, credential handli…
共 383283 条已上架资产
Design or review secure AgentMailer integrations, authorization boundaries, prompt-injection defenses, credential handli…
给 coding agent 设置和治理约束——决定一条规则该落到 hook、AGENTS
当用户搜索 VO3替代、Vovoo替代、对话视频Agent、prompt to video、视频修改、OiiOii同类平台、AI视频Agent、创意Agent平台、AI-HIVE MCP 时使用
Protect agent memory: snapshot before compaction, scan memory/snapshot files for prompt-injection, secrets, and contradi…
Talk to Loop (Möbius Loop) — the user's ranked queue of recurring tasks at mobiusprompt
ETF/行业投研平台的网关级前置意图识别与安全兜底Skill
Use when an agent wants to discover agent-made visual work, answer a shared visual-design prompt, or publish original pu…
元引 —— 意图澄清 + 生态入口技能:用户输入一句模糊的话 / 一个词时,识别意图、给出 2-4 个候选方向、深挖(目标/范围/输出/约束),再串联到对应元阁技能输出可直接运行的提示词
LLM 应用质量评测与回归测试实操手册——从"感觉不错"到"可度量可门禁":评测全景与指标体系(正确性/相关性/忠实度/幻觉率/鲁棒性/效率)、评测集构建(黄金数据集/对抗样本/领域评测集/规模估算)、RAG 系统评测(RAGAS 四指标:…
AI 模型与资产全生命周期管理(LLMOps)实操手册——从"有模型在用"到"资产清清楚楚":资产全景与台账(模型/Prompt/知识库/评估集/Agent 配置五类资产登记)、模型卡与注册(Model Card 字段规范、注册流程、元数据…
Convert the user's raw, unstructured, possibly disorganized thoughts (rambling, out-of-order, half-finished ideas) into …
帮用户生成去 AI 感的真人手持商品照片
本技能专为COS照片后期设计,能将用户模糊的修图需求(如加魔法阵、火焰、翅膀、塑形、风效等)转化为可直接导入Nano Banana的专业级JSON指令
多平台生成视频导演技能
元信 —— 装任何技能/包前的确定性安全扫描器:prompt injection(提示注入)+ 危险模式 + SKILL
Extend the Oracle-X financial terminal codebase — add or change a FastAPI endpoint, wire a new upstream data source into…
Run an iterative code-improvement hill-climb against real Datadog LLM-Obs data, locally, with Claude Code as the agent
Find every LLM workflow in the current repository and label it, so Caveman Cloud groups spend by what the code actually …
Helps AI Native PMs rapidly build runnable Prompt prototypes to validate product ideas
空间设计一人工作室的项目推进系统,旨在将模糊的客户需求转化为可讨论、可深化、可交付、可收款的设计框架与沟通文本