Skip to content

Translation in progress Some entries still include text in the original language.

Oct 12 · Monday

All papers

Found 2 papers
Filter5

Defense type

Instruction–data separation2Permissions and approval3

System type

AI Agent2coding agent0Multi-agent systems0

Attack surface / entry point

User input7Web pages and interfaces17Documents and messages10Tool responses24Tool and skill descriptions12Code repositories and dependencies2Memory writes5Inter-agent messages4Distribution channels5
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 240papers have no tags yet and are not included in filtered results.
Defense type：Instruction–data separationSystem type：AI AgentSystem type：coding agentSystem type：Multi-agent systemsAttack surface / entry point：Code repositories and dependenciesClear all

Another 240papers do not have category tags yet and are not included in filtered results.

  1. DefensesarXiv 2610.02569

    Pincer：用数字分身学习用户偏好并执行动态最小权限

    为 Agent 提出资源层授权系统 Pincer，用数字分身执行动态最小权限

    arXiv
    2610.02569
    Published
    Layer
    Application layer
    Setting
    AI Agent
    Evaluated models
    Claude Haiku 4.5, Muse Spark 1.2 Contributor
    AbstractPincer 在资源层以三票合取做授权。

    Pincer 是加在现成 coding agent 上的资源层授权防御：用只受用户侧信息影响的 digital twin 做动态最小权限，并与 tool-call 层防御并存。要处理的是用户维护策略带来的权限蠕变和授权疲劳，以及不学习用户策略的 tool-call 分类器。作者引述工作称 auto mode 会放行多数不安全操作，并认为推理代码或复杂 shell 的效果在这一层不实际。

    Paper analysisRead analysis
You have reached the end