菜单
Use when: changing Azure AI Foundry setup, model deployment validation, .agent_ids.json, agent staleness, create-agent, validate_setup.py, or supported GPT-5 model policy.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when: working with the Azure AI Foundry v2 SDK — AIProjectClient, agent creation (create_version / PromptAgentDefinition), the Responses API (responses.create), agent_reference vs agent, MCP/Web Search tools, streaming events, or azure-ai-projects version breakage.
Use when: changing PII checks, telemetry events, privacy boundaries, update checks, generated metrics, or docs/telemetry.md.
Use when: changing TSG Builder release workflow, PyInstaller build, Windows installer, artifact names, versioning, checksums, or release docs.
Use when: changing TSG Builder pipeline behavior, stage prompts, TSG markers, required headings, review warnings, retries, or Research/Write/Review contracts.
| name | foundry-agent-setup |
| description | Use when: changing Azure AI Foundry setup, model deployment validation, .agent_ids.json, agent staleness, create-agent, validate_setup.py, or supported GPT-5 model policy. |
Use this workflow for setup wizard, CLI validation, model policy, agent creation, and persisted agent metadata changes.
gpt-5.1, gpt-5.2, gpt-5.4, gpt-5.5.chat, mini, nano, pro, and codex variants, including versioned siblings.model_name for an existing deployment..agent_ids.json is v2-only and requires role objects with name, version, and id.model_deployment_name, underlying_model_name, and agent_definition_signature.model_policy.py, error_utils.py, web_app.py, and validate_setup.py.Run:
.venv/bin/pytest tests/test_model_policy.py tests/test_agent_staleness.py tests/test_web_endpoints.py::TestModelDeploymentValidation -q
make test-quick