From ba33469aa0d17393c9fb6d11aa3b7a165c3c8197 Mon Sep 17 00:00:00 2001 From: DB Lee Date: Thu, 14 May 2026 11:57:34 -0700 Subject: [PATCH] docs(http-agent): deduplicate judge env var block Re-validated docs/tutorial-http-agent.md against current develop (374 lines). One doc-only drift fixed. The Prerequisites block set both AZURE_OPENAI_DEPLOYMENT and AZURE_AI_MODEL_DEPLOYMENT_NAME to the same value ('gpt-4o-mini'). _model_config() reads them as fallbacks of each other - setting both is redundant. Reduced to one with a one-line alias note. Same fix shipped in PR #158 (#133) and #159 (#130). End-to-end verification (hybrid approach): * Section 2 az CLI: 'az containerapp up' on current 'containerapp' extension v1.3.0b4 still accepts every documented flag (--name, --resource-group, --location, --source, --target-port, --ingress external). * Sections 1, 3-6: ran the tutorial's exact FastAPI app.py locally at http://127.0.0.1:8000/chat (skipping the ACA deploy in Section 2) and pointed AgentOps at it. - 'agentops eval run' produced all six documented metrics (coherence, fluency, tool_call_accuracy, intent_resolution, task_adherence, avg_latency_seconds). - tool_calls extracted correctly via 'tool_calls_field: tool_calls' (row 0/1: tool_call_accuracy=5.00). Greeting row (no tools) shows tool_call_accuracy=n/a as expected. - 'results.json' and 'report.md' written under .agentops/results// and mirrored to /latest/. The ACA deploy itself wasn't exercised here, only the documented az CLI flag set; full ACA provisioning was deemed out of scope for doc-validation since AgentOps doesn't change behaviour based on ACA vs local HTTP URL. Refs #129. --- docs/tutorial-http-agent.md | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/docs/tutorial-http-agent.md b/docs/tutorial-http-agent.md index 80fd75a..4b67d2d 100644 --- a/docs/tutorial-http-agent.md +++ b/docs/tutorial-http-agent.md @@ -35,9 +35,12 @@ python -m pip install "agentops-toolkit[foundry,agent]" $env:AZURE_AI_FOUNDRY_PROJECT_ENDPOINT = "https://.services.ai.azure.com/api/projects/" $env:AZURE_OPENAI_ENDPOINT = "https://.openai.azure.com" $env:AZURE_OPENAI_DEPLOYMENT = "gpt-4o-mini" -$env:AZURE_AI_MODEL_DEPLOYMENT_NAME = "gpt-4o-mini" ``` +`AZURE_AI_MODEL_DEPLOYMENT_NAME` is accepted as a fallback name for the +judge deployment. Set only one of the two — `AZURE_OPENAI_DEPLOYMENT` +wins when both are set. + AgentOps is installed from the `develop` branch in this tutorial because the 1.0 tutorial surface is still being tested before the PyPI release.