Beyond prompt injection: a field guide to 20 open-source AI agent security tools
MaryamMiradi · x · 2026-11-07
The author surveyed the security concerns you must handle when building agents that browse the web, call tools, execute code, or touch internal systems — prompt injection, excessive permissions, unsafe tool use, leaked secrets, sensitive data exposure — and pulled 20 open-source security tools worth knowing, grouped into five categories:
- Prompt injection & guardrails: LLM Guard, NeMo Guardrails, AgentGuard, Guardian — for when untrusted text can influence agent behavior
- Red teaming & testing: PyRIT, garak, Promptfoo, Giskard — break the agent before someone else does
- Identity & permissions: Open Policy Agent, OpenFGA, Casbin, Cedar — control what users, agents, and tools can do
- Sandboxing & runtime: E2B, gVisor, Firecracker, Kata Containers — isolate execution with real side effects
- Data & secrets: Presidio, Trivy, Semgrep, ModelScan — sensitive data, credentials, and dependency scanning
Production agent security means sandboxing, permission enforcement, and data/secrets protection far beyond prompt injection; the author may benchmark these tools next.
More from coding & agent
- One folder that Claude reads first: the underrated Claude setup for solo operators — PrajwalTomar_ · 2026-11-07
- Linear's AI code review tool Reviews now free on all plans, no Linear tracking required — julianlehr · 2026-11-07
- Amazon blocks home-IP personal shopping agents; dev open-sources residential browser relay — unixterminal · 2026-11-07
- Stately ships a SKILL that has LLMs author agent state machines through 5 validation gates — DavidKPiano · 2026-11-07
- The AI agent testing landscape: 7 tools mapped across 5 distinct testing jobs — njyx · 2026-11-07
- AI rotoscoping workflow today: LTX 2.5 Alpha LoRA + SAM3 + BiRefNet, layered ProRes — PurzBeats · 2026-11-07