asrt-bench
PulseGate's liveness check found it on 21 Sep 2026; it is registered on GitHub and PyPI and has been in the index since 21 Sep 2026. How this is checked
asrt-bench is an MIT-licensed command-line benchmark for evaluating AI agents against a frozen attack pack. It verifies attack outcomes from tool-call traces and compares versions to identify safety regressions.
Inferred · not functionally tested
Overview
5 featuresPurpose: Testing AI agents for prompt-injection vulnerabilities and detecting safety regressions across versions.
Inferred · not functionally tested
Audience: AI security engineers and developers building AI agents
Inferred · not functionally tested
Functions: agents
Inferred · not functionally tested
Interfaces: API: unknown · MCP: unknown · 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.
asrt-bench is an Agent evaluation & testing project. Inferred · not functionally tested: It focuses on testing AI agents for prompt-injection vulnerabilities and detecting safety regressions across versions. Inferred · not functionally tested: It is built as an open-source project for AI security engineers and developers building AI agents. Basis unknown · not verified: asrt-bench is open source under the MIT license. Basis unknown · not verified: asrt-bench is available on the command line, and it can be self-hosted.
Behind asrt-bench is m4vic, and it first shipped in 2026. The project is developed in the open on GitHub with 19 commits in the last 90 days. Inferred · not functionally tested: Key capabilities include frozen attack packs, prompt injection tests, and tool-call verification.
Summary written by a language model from the project’s public pages.
Tasks: Inferred · not functionally tested
- Frozen attack packs
- Prompt injection tests
- Tool-call verification
- Version diffing
- Safety regression detection
Topics: Inferred · not functionally tested
Built with & integrations
- Claude Code
- commit e454ceb4a14a · since Aug 2026
Trust & compliance
Indexing history
2What PulseGate has recorded for this listing
- Indexed21 Sep · 17:38 UTCasrt-bench seen via PyPI Bulk EnumeratorSource: PyPI Bulk Enumerator · Open
Frequently asked questions about asrt-bench
- What does asrt-bench do?
- Inferred · not functionally tested: Asrt-bench focuses on testing AI agents for prompt-injection vulnerabilities and detecting safety regressions across versions. It is catalogued under Agent evaluation & testing on PulseGate.
- Who is asrt-bench for?
- Inferred · not functionally tested: asrt-bench is an open-source project built for AI security engineers and developers building AI agents.
- Does asrt-bench have a free plan?
- Basis unknown · not verified: Yes — asrt-bench is open source under the MIT license and free to use.
- What platforms does asrt-bench run on?
- Basis unknown · not verified: asrt-bench runs on the command line. It can also be self-hosted.
- Is asrt-bench still active?
- PulseGate's liveness check found it on 21 Sep 2026. Its GitHub repository shows 19 commits in the last 90 days.
- What are alternatives to asrt-bench?
- Similar projects tracked by PulseGate include agentbench-cli, ssebench, and agent-security-bench.agentbench-clissebenchagent-security-bench
- Who makes asrt-bench?
- asrt-bench is developed by m4vic.
- When did asrt-bench launch?
- asrt-bench first shipped in 2026.
Similar projects
Closest matches by what these projects do