agent-security-bench
PulseGate's liveness check found it on 3 Oct 2026; it is registered on GitHub and PyPI and has been in the index since 16 Aug 2026. How this is checked
agent-security-bench is an open-source Python benchmark for evaluating ML coding agents and agent security. It provides weighted checks, AST-based scoring, security tests such as prompt-injection and SSRF checks, and machine-readable evaluation receipts.
Inferred · not functionally tested
Overview
6 featuresPurpose: Evaluating coding-agent behavior and security risks consistently with automated, weighted, machine-readable checks.
Inferred · not functionally tested
Audience: AI security researchers and developers evaluating coding agents
Inferred · not functionally tested
Functions: analytics
Inferred · not functionally tested
Interfaces: API: unknown · MCP: indicated (inferred, not tested) · CLI: indicated (inferred, not tested) · Self-hosting: indicated (inferred, not tested)
Recorded constraints: pricing: open_source · license: MIT · platforms: CLI · deployment: cli, self_hosted
Constraint provenance is unknown; confirm requirements with the publisher.
Record sources: pypi.org · github.com. These links do not verify the individual claims.
agent-security-bench is an Agent evaluation & testing project. Inferred · not functionally tested: It focuses on evaluating coding-agent behavior and security risks consistently with automated, weighted, machine-readable checks. Inferred · not functionally tested: It is built as an open-source project for AI security researchers and developers evaluating coding agents. Basis unknown · not verified: agent-security-bench is open source under the MIT license. Basis unknown · not verified: agent-security-bench is available on the command line, and it can be self-hosted.
Behind agent-security-bench is Sina Kazemnezhad, and it first shipped in 2026. The project is developed in the open on GitHub with 4 commits in the last 90 days. Inferred · not functionally tested: Among its 6 catalogued features are weighted checks, AST scoring, and security evaluation. Basis unknown · not verified: Catalogued interfaces include an MCP server.
Summary written by a language model from the project’s public pages.
Tasks: Inferred · not functionally tested
- Weighted checks
- AST scoring
- Security evaluation
- Prompt-injection testing
- Jailbreak testing
- SSRF checks
Topics: Inferred · not functionally tested
Built with & integrations
Trust & compliance
Indexing history
2What PulseGate has recorded for this listing
Frequently asked questions about agent-security-bench
- What is agent-security-bench?
- Inferred · not functionally tested: Agent-security-bench focuses on evaluating coding-agent behavior and security risks consistently with automated, weighted, machine-readable checks. It is catalogued under Agent evaluation & testing on PulseGate.
- Who should use agent-security-bench?
- Inferred · not functionally tested: agent-security-bench is an open-source project built for AI security researchers and developers evaluating coding agents.
- Does agent-security-bench have a free plan?
- Basis unknown · not verified: Yes — agent-security-bench is open source under the MIT license and free to use.
- What platforms does agent-security-bench run on?
- Basis unknown · not verified: agent-security-bench runs on the command line. It can also be self-hosted.
- Is agent-security-bench still active?
- PulseGate's liveness check found it on 3 Oct 2026. Its GitHub repository shows 4 commits in the last 90 days.
- What projects are similar to agent-security-bench?
- Similar projects tracked by PulseGate include agent-evaluation-lab, agentbench-cli, and agent-safety-bench-envs.agent-evaluation-labagentbench-cliagent-safety-bench-envs
- Who makes agent-security-bench?
- agent-security-bench is developed by Sina Kazemnezhad.
- How long has agent-security-bench been around?
- agent-security-bench first shipped in 2026.
Similar projects
Closest matches by what these projects do