PinnedSrinivas·Oct 20Testing Agentic AI, Week 1: why an agent that is 80% right is 100% wrongThis is the second article in my 21-part Medium series on testing Agentic AI. The map of the whole series is in the overview article, "A…
Srinivas·3h agoTesting Agentic AI, Week 6: Autonomy Needs LimitsEnvironment changes, budgets, time limits, handoffs and escalation rules you can actually test: posts 26 to 30 of…
Srinivas·22h agoAiEnsured agentest Hands-On: Your First Trace CheckOne pip install, one command, and a shopping agent that said “added to the cart” when the cart was empty. Then the Colab, the 56-second…
Srinivas·22h agoTesting Agentic AI, Week 5: Handoffs, Delegation, RollbackConstraints that survive a handoff, delegation that is not permission, sent versus accepted subgoals, dependency ordering and rollback…
Srinivas·6d agoI’m launching agentest: test what your agent did, not only what it saidAn open suite of 130 executable checks for deterministic agent red-teaming and release gating. Clone the repo and install from source.
Srinivas·6d agoTest AI Agents for Unauthorized Tool Calls in 5 MinutesHow to test AI agents for undeclared tool calls and off-list URLs, offline, from OpenTelemetry traces. Open source, no LLM judge, no API…
Srinivas·Nov 15Prompt Injection Testing for AI Agents, Using TracesHow to test an AI agent for prompt injection from recorded traces: planted canaries, recipient and host allowlists, and a clean versus…
Srinivas·Nov 9Testing Agentic AI, Week 4: how do you test an agent’s defenses?Week 4 of the 100-day series: continuous red-teaming, queryability, audit trails, teaching validation, and least privilege (posts 16 to 20)
Srinivas·Nov 3Testing Agentic AI, Week 3: Handoffs, Order and PrivacyExplaining LLM answers, handoffs, task order, agency and DPDP privacy: posts 11 to 15 of #100daysoftestingagenticai