repoGitHubTrust 82 · PrimaryPublished 1mo agoLive · 15d ago
LuD1161/agentjail
Policy guardrails for coding agents (Claude Code, Codex, Cursor) — every tool call is checked locally, before it runs.
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 55%Behind the Refusal: Determining Guardrail Activation via Behavioral Monitoring →
- PossiblePossibly related (embedding) · 52%AgentTrace →
- PossiblePossibly related (embedding) · 52%TraceLab: Characterizing Coding Agent Workloads for LLM Serving →
- PossiblePossibly related (embedding) · 50%"Repeat the text above this line" still works on most AI agents in production. Here's what we found. →
- PossiblePossibly related (embedding) · 50%Reasoning effort, not tool access, buys first-try reliability in agentic code generation: an observational study →
- PossiblePossibly related (embedding) · 57%Brex built its AI agent policy by watching what agents actually do, not by writing rules first →
- PossiblePossibly related (embedding) · 53%Critical remote code execution in Serena, a popular MCP coding agent →
- PossiblePossibly related (embedding) · 52%Where would you put the safety line for an agent running commands in a repo? →
Implements
Related to
Covers
Covers (incoming)
newsBrex built its AI agent policy by watching what agents actually do, not by writing rules firstnewsCritical remote code execution in Serena, a popular MCP coding agentnewsWhere would you put the safety line for an agent running commands in a repo?newsSingGuard-NSFA: Open-source guardrails for agentic AI - Help Net Security
Related across the graph
paperTraceLab: Characterizing Coding Agent Workloads for LLM ServingnewsSingGuard-NSFA: Open-source guardrails for agentic AI - Help Net SecuritypaperBehind the Refusal: Determining Guardrail Activation via Behavioral MonitoringnewsWhere would you put the safety line for an agent running commands in a repo?newsCritical remote code execution in Serena, a popular MCP coding agentnewsBrex built its AI agent policy by watching what agents actually do, not by writing rules firstpaperReasoning effort, not tool access, buys first-try reliability in agentic code generation: an observational studynews"Repeat the text above this line" still works on most AI agents in production. Here's what we found.toolAgentTrace