Codex Knowledge Base
Model Routing Patterns for Coding Agents: A Sol, Terra, Luna Decision Framework
Model Routing Patterns for Coding Agents: A Sol, Terra, Luna Decision Framework
The Friendly Fire Exploit: How Defensive Code Review Becomes Remote Code Execution — and Why Codex CLI's Default Posture Blocks the Kill Chain
The Friendly Fire Exploit: How Defensive Code Review Becomes Remote Code Execution — and Why Codex CLI’s Default Posture Blocks the Kill Chain
Do Auto-Generated AGENTS.md Files Actually Help? What Three Studies Say About Instruction File Effectiveness
Do Auto-Generated AGENTS.md Files Actually Help? What Three Studies Say About Instruction File Effectiveness
Cross-Agent Context Transfer: Why Interop Is the Next Bottleneck for Coding Agents
Cross-Agent Context Transfer: Why Interop Is the Next Bottleneck for Coding Agents
Nine Outages in Seven Days: Building Codex CLI Resilience with Multi-Provider Failover
Nine Outages in Seven Days: Building Codex CLI Resilience with Multi-Provider Failover
GuardFall and ShellSieve: Why Denylist-Based Security Is a Losing Strategy for Coding Agents — and How Codex CLI's OS-Level Sandbox Changes the Calculus
GuardFall and ShellSieve: Why Denylist-Based Security Is a Losing Strategy for Coding Agents — and How Codex CLI’s OS-Level Sandbox Changes the Calculus
Outcome-First Prompting: How GPT-5.6 Sol's Lean-Prompt Philosophy Reshapes Your Codex CLI Strategy
Outcome-First Prompting: How GPT-5.6 Sol’s Lean-Prompt Philosophy Reshapes Your Codex CLI Strategy
The GPT-5.6 Serialisation Tax: How Explicit Batching in Code Mode Cuts Your Codex CLI Token Bill by 27–45%
GPT-5.6 silently serialises independent tool calls in Code Mode instead of batching them. A community-discovered AGENTS.md directive using Promise.allSettled() reduced weighted usage by 27–45% and model cycles by over 50%....
Do Auto-Generated AGENTS.md Files Actually Help? What Four Studies Reveal About /init, Context Bloat, and the Real Cost of Repository Guidance
Do Auto-Generated AGENTS.md Files Actually Help? What Four Studies Reveal About /init, Context Bloat, and the Real Cost of Repository Guidance
Inside v0.145's Terminal Rendering Pipeline: How Incremental Markdown, Inline Visualisation Links, and Bounded Output Fixed Codex CLI's Long-Session Performance
Inside v0.145’s Terminal Rendering Pipeline: How Incremental Markdown, Inline Visualisation Links, and Bounded Output Fixed Codex CLI’s Long-Session Performance