CODEX FIELD MANUAL · AUTONOMOUS PLAYBOOKS

Codex Survival Playbooks & Feature Guides

Zero-fluff tactical guides on OpenAI Codex quota optimization, AGENTS.md memory, parallel subagents, prompt caching, and live @thsottiaux feature breakdowns — auto-synthesized in 5 languages.

🤖 Event-Driven Autonomous Pipeline · 5 Locales 📚 9 Playbooks 🤖 /llms.txt

⚡ Models & Caching ⏱ 3 min read ⚡ AI Radar

25 hours ago

Codex Ship Breakdown: OpenAI Codex CLI rust-v0.165.0 (Added a built-in Google Cloud Vertex AI prov)

Tactical breakdown and quota impact analysis for "OpenAI Codex CLI rust-v0.165.0 (Added a built-in Google Cloud Vertex AI prov)" (github.com/openai/codex (rust-v0.165.0)).

TL;DR

## New Features - Added a built-in Google Cloud Vertex AI provider, configured with `model_provider = "gcp"`. (#56265) - Added GPT-6.1 Terra to the model catalogs. (#57337) - Adde — High Efficiency — Prefer the newly shipped model tier for daily coding to maximize effective quota per reset window.

Signal Source: github.com/openai/codex (rust-v0.165.0)Category: MODELSQuota Impact: Positive (Saves Quota)

👀 120👍 24 .md

Read Playbook → →

📋 AGENTS.md & Workflow ⏱ 4 min read ⚡ AI Radar

31 hours ago

Codex CLI Automation, Statusline & MCP Cookbook: Build a Zero-429 Terminal Workflow

Wire live Codex reset telemetry directly into your tmux status bar, AI coding agent MCP config, and Feishu/Discord/Slack team webhooks.

TL;DR

Connect npx tibosay --mcp to your coding agent so it automatically checks live reset probability before running token-heavy tasks, and embed /api/codex/ansi?mode=statusline in tmux.

MCP Protocol: stdio JSON-RPC (npx -y tibosay@latest --mcp)Statusline Endpoint: GET /api/codex/ansi?mode=statuslineSupported Team Bots: Feishu, WeCom, Discord, Slack, JSON

👀 120👍 24 .md

Read Playbook → →

⚡ Models & Caching ⏱ 4 min read ⚡ AI Radar

37 hours ago

GPT-6.1 Terra vs. Sol in Codex: How to Unlock 2X Usage Multiplier & Computer-Use Workflows

Tibo announced GPT-6.1 Terra on Day 22 with 2X more usage than 6.1 Sol across API and subscriptions. Here is when to use Terra vs. Sol to double your effective coding time.

TL;DR

GPT-6.1 Terra gives 2X more usage quota than GPT-6.1 Sol across both subscriptions and API while excelling at computer use, UI implementation, and nuanced instruction following. Make Terra your default daily driver.

GPT-6.1 Terra Quota Multiplier: 2X more usage vs. GPT-6.1 SolCore Strengths: Computer Use, Nuanced Requests, Full-StackRecommended Task Share: 85% Terra / 15% Sol

👀 120👍 24 .md

Read Playbook → →

⚡ Models & Caching ⏱ 4 min read ⚡ AI Radar

43 hours ago

Mastering Codex 1M Context Window & Prompt Caching: Maximize Cache Hit Rates in Long Sessions

How Codex CLI leverages persistent WebSockets and prefix prompt caching across 1M context tokens — and the 4 habits that keep your cache hit rate above 90%.

TL;DR

Prompt caching discounts repeated prefix tokens by up to 75–90%, but only if the prefix stays byte-identical. Keep AGENTS.md static during a session, avoid switching models mid-thread, and compact before hitting context cliffs.

Max Context Window: 1,000,000 Tokens (1M)Transport Protocol: Persistent Low-Latency WebSocketsWarm Cache Savings: Up to 75%–90% lower prefix cost

👀 120👍 24 .md

Read Playbook → →

🐝 Subagents & Sandbox ⏱ 5 min read ⚡ AI Radar

2 days ago

Parallel Subagents & Git Worktrees in Codex: High-Velocity Multi-Agent Coding Without Quota Drain

Learn when to fan out parallel Codex subagents, how to isolate concurrent tasks with Git worktrees and bubblewrap sandboxes, and how to avoid N-times quota multiplication.

TL;DR

Use a read-only Research Subagent to scan large folders without polluting your main context window, and pair parallel write agents with isolated Git worktrees — but only fan out 3+ write agents when your quota is freshly reset or in a High Watch burn window.

Research Subagent Impact: Saves 30%+ main-thread context tokensParallel Write Agents: Linearly multiplies token burn (N×)Workspace Isolation: git worktree + bubblewrap sandbox

👀 120👍 24 .md

Read Playbook → →

📋 AGENTS.md & Workflow ⏱ 5 min read ⚡ AI Radar

2 days ago

AGENTS.md Engineering Playbook: Cut Codex Token Burn by 40% & Eliminate Retry Loops

How to structure root and nested AGENTS.md files so OpenAI Codex finds the right files, runs the exact test command on the first try, and stops burning quota on failed guesses.

TL;DR

The #1 cause of Codex quota drain is blind repo exploration and failed test/linter invocations. A 40-line AGENTS.md with exact single-file test commands, directory boundaries, and architecture invariants cuts token consumption by 35–45%.

Average Quota Saved: 35% – 45% fewer tokens per taskIdeal Root File Length: 40 – 80 lines (Prompt Cache friendly)Nested Discovery: Automatic root -> cwd inheritance

👀 120👍 24 .md

Read Playbook → →

🪙 429 & Quota Strategy ⏱ 4 min read ⚡ AI Radar

3 days ago

The Complete Codex 429 Survival Guide: Regular Resets vs. Banked Resets Strategy

Understand how OpenAI Codex usage windows work, why global resets overwrite unused regular quota, and when to trigger your Banked Reset without wasting tokens.

TL;DR

Global Regular Resets restore everyone to 100% and overwrite unused regular quota — burn remaining regular quota when TiboSay Oracle exceeds 75%, and save Banked Resets strictly for urgent 429 blocks during low-probability windows.

Regular Reset Carryover: None (Overwrites to 100%)Banked Reset Stacking: Yes (Manual activation on 429)Golden Drop Window: 10:00 AM – 9:00 PM PT (84%)

👀 120👍 24 .md

Read Playbook → →

⚡ Models & Caching ⏱ 3 min read ⚡ AI Radar

5 days ago

Day 19 Breakdown: Dots calls now connect much faster

Tactical breakdown and quota impact analysis for "Dots calls now connect much faster" (@thsottiaux X Feature Ship).

TL;DR

Dots calls now connect much faster. Changing reasoning mid-conversation in Codex and Work no longer invalidates the cache. — High Efficiency — Prefer the newly shipped model tier for daily coding to maximize effective quota per reset window.

Signal Source: @thsottiaux X Feature ShipCategory: MODELSQuota Impact: Positive (Saves Quota)

👀 120👍 24 .md

Read Playbook → →

⚡ Models & Caching ⏱ 3 min read ⚡ AI Radar

5 days ago

Day 19 Breakdown: Codex CLI now supports reusable prompt presets, inline model and per

Tactical breakdown and quota impact analysis for "Codex CLI now supports reusable prompt presets, inline model and per" (@thsottiaux X Feature Ship).

TL;DR

Codex CLI now supports reusable prompt presets, inline model and permission settings, enterprise MCP sign-in, and Windows corporate-proxy authentication. Failures when resuming lon — High Efficiency — Prefer the newly shipped model tier for daily coding to maximize effective quota per reset window.

Signal Source: @thsottiaux X Feature ShipCategory: MODELSQuota Impact: Positive (Saves Quota)

👀 120👍 24 .md

Read Playbook → →