Agentic AI testing and assurance — building open tools and checklists so teams can judge AI agents with evidence, not vibes.
CTO & Co-founder, AiEnsured / testAIng.com. PhD 2000, University of Alberta. Upcoming book on Agentic AI.
Contact: srinivas AT testAIng.com
| agentest (offline agent test runner) | https://github.com/srpadmanabhuni/agentest |
| Agentic AI Test Checklist 110 (Hugging Face) | https://huggingface.co/datasets/srinivaspadmanabhuni/agentic-ai-test-checklist-110 |
| Assurance Report Card (HF Space) | https://huggingface.co/spaces/srinivaspadmanabhuni/agentest-assurance-report-card |
| Boardroom AI Brief (Substack) | https://boardroomaibrief.substack.com |
| Site | https://sites.google.com/view/srinivaspadmanabhuni |
| ORCID | https://orcid.org/0009-0002-3904-7204 |
agentest is an offline, deterministic test runner for agentic AI. It covers the open Agentic AI Test Checklist 110: give it an agent's OpenTelemetry GenAI traces (plus optional A2A AgentCard and MCP tools) and it returns pass / fail / n/a for each check. No account, token, network, or model judge required.
- Try the Assurance Report Card Space
- Pick a
good first issueif you want to contribute - Subscribe to Boardroom AI Brief for short notes on AI governance and assurance
pip install git+https://github.com/srpadmanabhuni/agentest
agentest std run --agent a14