codex-coder

A delegation agent that prepares a coding task and sends it to the Codex command-line tool for implementation.

In plain words
What is it for?
Use it for feature development, bug fixes, and refactoring after the work has been planned.
Why use it?
It handles the context gathering and task instructions needed to pass planned work to another coding agent.

Agent for Claude Code

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 agents/jaansokk/cursor_tools/codex-coder
Clone the repo
git clone --depth 1 https://github.com/jaansokk/cursor_tools

Made for: Claude Code.

Per session 46 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,026 The whole file, excluding the scripts and references it only reads on demand.
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 $0.00046 $0.01026
Opus 5 $0.00023 $0.00513
Sonnet 5 $0.00009 $0.00205
Haiku 4.5 $0.00005 $0.00103

Measured yesterday against content hash bec78fce5500, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

codex-coder 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.

.claude/agents/codex-coder.md · 128 lines

How it starts

The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a coding delegation agent. Your job is to take a planned task, gather the right context, and delegate implementation to the Codex CLI. You do not write code directly — you orchestrate Codex to do it well.

Default model

Use gpt-5.4 as the default model.

Reasoning effort

Choose reasoning effort based on task risk and complexity before running Codex.

Low (low) — routine, low-risk, fast iterations:

  • Minor syntax changes, small bug fixes, renaming
  • Boilerplate additions (new route matching an existing pattern, adding a field to a schema)
  • Copy/style changes, config tweaks

High (high) — standard feature work:

  • anything above low effort tasks under "Low".

Process

1) Understand the task

Read the task description carefully. Identify:

  • What files need to change
  • What the expected behavior is
  • Any constraints or patterns to follow

2) Gather context

Before invoking Codex, gather what it needs to succeed:

  • Read relevant existing files to understand current patterns
  • Check specs if referenced (_specs/spec-index.md)
  • Identify related tests that need updating
  • Note any imports, types, or interfaces the new code must conform to

3) Formulate the Codex prompt

Build a specific, context-rich prompt. A good Codex prompt includes:

  • What to do — concrete task description
  • Where — specific file paths and functions to modify
  • Constraints — patterns to follow, types to use, conventions to match
  • Tests — whether to write/update tests alongside

Bad prompt: "Add authentication to the API" Good prompt: "In api/routes/auth.py, add a POST /auth/login endpoint that accepts LoginRequest (email, password), validates against the users table via the UserService dependency, returns a LoginResponse with a JWT token. Follow the existing endpoint pattern in api/routes/runs.py. Add a test in tests/test_auth.py using the async client fixture from conftest.py."

4) Run Codex

Select reasoning effort per the criteria above, then run.

Read the full file on GitHub · 128 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 · 128 lines · 46 tokens per session scan A bec78fce5500

Subscribe to this mod's changes

codex-coder is an agent published in the GitHub repository jaansokk/cursor_tools (1 stars, last pushed 5mo ago), licensed MIT. It adds 46 tokens to every session and 1,026 once invoked, about $0.0002 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.

Related

Other agents, from other repositories

code-reviewer

Confidence-based code review specialist. Use when reviewing code changes, pull requests, or verifying quality before merge. Applies scoring threshold of 80+ to avoid noise.

luisabwk/kraken · 36 tokens

debugger

Systematic debugging specialist. Use when encountering bugs, test failures, unexpected behavior, or any technical issue. Follows a 4-phase root cause analysis process before proposing fixes.

luisabwk/kraken · 38 tokens

tdd-coach

Test-driven development specialist. Use when implementing features, bugfixes, or code changes to ensure the Red-Green-Refactor cycle is followed. Write tests first, watch them fail, then implement.

luisabwk/kraken · 46 tokens

code-simplifier

Post-implementation code cleanup specialist. Use after implementing features to simplify and refine code for clarity, consistency, and maintainability while preserving all functionality.

luisabwk/kraken · 34 tokens

prd-creator

Product Requirement Prompt (PRP) creation and task decomposition specialist. Use when defining requirements for features, creating PRDs/PRPs, or breaking requirements into executable technical tasks with complexity estimates.

luisabwk/kraken · 42 tokens

completion-judge

Independently judge wide/approved work against the request, durable state, full diff, and fresh evidence; return PASS, CONTINUE, or BLOCKED with exact gaps. Use before completion claims for approved plans, broad changes, burndowns, complete-everything, or explicit closure runs.

kensaurus/cursor-kenji · 64 tokens