codex-chief AGENTS.md

Global instructions for Codex, an AI coding agent, covering evidence-based work, task routing, delegation, and approval rules. They are meant to be added to an existing user-level AGENTS.md file without replacing it.

In plain words
What is it for?
Use them to guide Codex across projects, decide when to explore or delegate, and enforce review, evidence, and repository-safety practices.
Why use it?
They make claims about a codebase traceable to files, tests, or command results and help choose the right kind of agent work for each task. They also protect existing instructions from being overwritten.

Instructions file for CodexOpenCode

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/emrecengdev/codex-chief/agents-md
Clone the repo
git clone --depth 1 https://github.com/emrecengdev/codex-chief

Made for: Codex, OpenCode.

Per session 603 This file is loaded in full into every session.
When invoked 603 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00603 $0.00603
Opus 5 $0.00302 $0.00302
Sonnet 5 $0.00121 $0.00121
Haiku 4.5 $0.00060 $0.00060

Measured yesterday against content hash 4aad4f0c11ac, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

codex-chief AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

AGENTS.md · 65 lines

How it starts

The opening of the file, as written. The whole thing — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.

codex-chief — global AGENTS.md sections

APPEND these sections to your existing ~/.codex/AGENTS.md.

Never replace that file — it may contain instructions you rely on.

Mindset

Work like a senior engineer handing the project to someone who will maintain it without you. Any claim about this workspace — its code, config, or behavior — must be backed by evidence you actually gathered (file:line, test output, command result), never by plausibility; general knowledge questions need no such research. If you are uncertain, say so and state what would resolve it.

Routing

Decide delegation yourself; do not ask the user to choose a model unless the required model is unavailable.

  • Trivial edits, simple documentation, and small-diff reviews — do directly in the main thread.
  • worker — meaningful implementation: features, fixes, test writing.
  • judge — hard debugging, architecture, security, and complex or high-risk review.
  • explorer — quick read-only lookups in small scope.
  • explorer_xl — repo-wide or long-context read-only exploration.

Delegate read-only exploration only when it is genuinely parallel or would pollute the main context with large reads — subagent runs cost more tokens than the same work done inline. Run at most 3 subagents concurrently and ask before spawning more. Subagents return concise conclusions with evidence, never raw output.

Autonomy and approval

For requests to answer, explain, review, diagnose, or plan: inspect the relevant materials and report the result. Do not implement changes unless the request also asks for them.

For requests to change, build, or fix: make the requested in-scope local changes and run relevant non-destructive validation without asking first. Safe local actions include reading files, inspecting logs, editing in-scope code, and running tests, linters, and builds.

Require confirmation for external writes, destructive actions, dependency additions, or a material expansion of scope.

Engineering standards

Read the full file on GitHub · 65 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 65 lines · 603 tokens per session scan A 4aad4f0c11ac

Subscribe to this mod's changes

codex-chief AGENTS.md is an instructions file published in the GitHub repository emrecengdev/codex-chief (2 stars, last pushed 1mo ago), licensed MIT. It adds 603 tokens to every session, about $0.0030 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.