Official Technical Resources: OpenAI (page 15)
Official Technical Resources · OpenAI: 258 entries in total; this page lists entries 211–225.
Developer notes on the Realtime API
Details worth noticing in recent realtime speech-to-speech updates
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Speech Recognition and Synthesis · Streaming Output
Link last checked: 2026-09-24
Making private MCP servers reachable without making them public
How we preserved private network boundaries while supporting MCP streaming, authentication, and an inspectable client.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: AI Guardrails · Agent Identity, Authorization, and Secrets
Link last checked: 2026-09-24
Rethinking skills and prompts for GPT-6 Astra
Revisit skill descriptions, AGENTS.md, and task prompts to avoid bloated context.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Model Context Protocol (MCP) · Tool Calling
Link last checked: 2026-09-24
Run long horizon tasks with Codex
Kept as a technical article for its implementation value but placed in the supplementary tier by default; core behavior and long-term contracts follow the main guides and formal references.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Agent Memory · Context Compaction
Link last checked: 2026-09-24
Shell + Skills + Compaction: Tips for long-running agents that do real work
Practical patterns for building with skills, hosted shell, and server-side compaction in the Responses API.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Agent Memory · Context Compaction
Link last checked: 2026-09-24
Testing Agent Skills Systematically with Evals
A practical guide to turning agent skills into something you can test, score, and improve over time.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Model Evaluation and Benchmarks · LLM Application Evaluation
Link last checked: 2026-09-24
Using skills to accelerate OSS maintenance
Using skills and GitHub Actions to optimize Codex workflows in the OpenAI Agents SDK repos.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Model Context Protocol (MCP) · Tool Calling
Link last checked: 2026-09-24
What makes a great ChatGPT app
How to build capabilities that make conversations better.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Workflow Orchestration · Model Deployment Options
Link last checked: 2026-09-24
Why we built the Responses API
How the Responses API unlocks persistent reasoning, hosted tools, and multimodal workflows for GPT-5.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Model Context Protocol (MCP) · Tool Calling
Link last checked: 2026-09-24
Build a per-run spending controller with the Responses API
Enforce an application-owned model-token budget for each Responses API run using token counts, maximum-cost estimates, usage tracking, and safe error handling.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Agent Loop · Structured Outputs
Link last checked: 2026-09-24
OpenAI Harmony Response Format
Kept as an engineering recipe for its implementation value but placed in the supplementary tier by default; core behavior and long-term contracts follow the main guides and formal references.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Agent Loop · Structured Outputs
Link last checked: 2026-09-24
User guide for gpt-oss-safeguard
Kept as an engineering recipe for its implementation value but placed in the supplementary tier by default; core behavior and long-term contracts follow the main guides and formal references.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: AI Guardrails · Agent Identity, Authorization, and Secrets
Link last checked: 2026-09-24
Verifying gpt-oss implementations
Kept as an engineering recipe for its implementation value but placed in the supplementary tier by default; core behavior and long-term contracts follow the main guides and formal references.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Workflow Orchestration · Model Deployment Options
Link last checked: 2026-09-24
AML Analysis with the Agents SDK on Amazon Bedrock
Run a synthetic AML analysis with the OpenAI Agents SDK and GPT-5.6 Sol on Amazon Bedrock, verify structured claims against deterministic evidence, test failure cases, and enforce an application-owned human review gate.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: AI Guardrails · Agent Identity, Authorization, and Secrets
Link last checked: 2026-09-24
Build a data analyst with the Agents API
Investigate warehouse questions with read-only SQL, business context, and saved analyst corrections.
Evidence use: For understanding, implementation, selection, reliability, security or migration decisions.
Limitations: Directly evidences only OpenAI's own implementation and claims. Recheck the page status and version before use.
Related concepts: Workflow Orchestration · Model Deployment Options
Link last checked: 2026-09-24