agent-evaluations

Episodes (4)

Trajectory notes

• 2026-08-22T16:30:01Z: tracelint-deterministic-agent-trace-checks closed (faded) — Scott already holds this exact position in `path-testing`: deterministic regression checks over agent traces, including tool-call sequence and schema assertions, with repeatable gates rather than relying solely on LLM
• 2026-08-20T16:41:59Z: agentgauntlet-failure-benchmark closed (faded) — Reflexive Agent Design and Trace-backed agent comparison already establish Scott’s position that agent systems should be evaluated through reproducible, path-level traces under varied conditions rather than outcomes alone. AgentG
• 2026-08-19T01:23:56Z: oqoqo-agent-interface-evals closed (faded) — Oqoqo productizes the evaluation loop already described in `Reflexive Agent Design` and `Evaluation-Driven Development`: real agent traffic over MCP/CLI surfaces, retained trajectories, repeatable regressions, and CI-triggered qualit