agentest 3.5 lands on PyPI for offline AI agent testing
The free, open-source agentest tool has launched version 3.5 on PyPI as agentest-ai, giving teams a deterministic way to test what AI agents actually do from recorded traces. The release matters because it lets QA teams, developers and researchers verify agent behavior offline without API keys, network access or LLM judges.
Why it matters: - AI agents can produce a correct-looking answer while still taking the wrong action. - agentest checks recorded agent traces to verify what the agent actually did, not only what it said. - The tool is designed for QA teams, developers and researchers working with tool-using AI agents. - Each test result includes an evidence line that teams can keep for records.
What happened: - agentest version 3.5 released on November 19, 2026. - The release is the project’s first publication on the Python Package Index, where it is available as the agentest-ai package. - The tool is free and open source under the Apache-2.0 license. - The project is created and maintained by Dr. Srinivas Padmanabhuni, CTO and co-founder of testAIng.com and an AI researcher based in Bengaluru, India.
The details: - agentest runs deterministic offline tests. - The tool does not require an API key. - The tool does not need network access. - The tool does not use an LLM judge. - agentest reads traces that an agent already records and evaluates the agent’s actions from that evidence. - The project is the automated companion to the Agentic AI Test Checklist. - The checklist is a free catalogue of 110 test designs published on Hugging Face. - The checklist defines what to test in an AI agent and what counts as passing. - Teams can use the checklist on its own for test planning, manual reviews and evidence records. - The project page is available here. - The checklist is available here.
Between the lines: - The release reflects a push toward repeatable agent testing that does not depend on live model calls or external review systems. - An offline, evidence-based approach can make agent QA easier to audit and reproduce. - Positioning the checklist and the software together suggests the project is aiming to cover both test design and test execution.
What's next: - Teams can install agentest-ai from PyPI and use it for offline validation of agent behavior. - Developers and QA teams can pair the tool with the checklist to plan tests and capture evidence. - The project is likely to serve users who need reproducible checks for agents that call tools and take actions.
Disclaimer: This article was produced by AGP Wire with the assistance of artificial intelligence based on original source content and has been refined to improve clarity, structure, and readability. This content is provided on an “as is” basis. While care has been taken in its preparation, it may contain inaccuracies or omissions, and readers should consult the original source and independently verify key information where appropriate. This content is for informational purposes only and does not constitute legal, financial, investment, or other professional advice.
Sign up for:
Delhi Daily Times
The daily local news briefing you can trust. Every day. Subscribe now.
Check Your Email!
We sent a one-time activation link to: .
Confirm it's you by clicking the email link.
If the email is not in your inbox, check spam or try again.
Welcome back!
is already signed up. Check your inbox for updates.