delegating-to-codex

delegating-to-codex is a skill for Claude Code, Codex from isvlasov/rageatc-oss. It costs 103 tokens per session (1,020 once invoked), scanned B, original, MIT.

A way to hand coding tasks to a separate OpenAI Codex command-line session that stays visible in a herdr terminal pane. You can guide the session while it works and review its results.

In plain words
What is it for?
Use it to delegate implementation, ask for a second opinion on design or reasoning, or run lighter and harder tasks with different Codex models.
Why use it?
It lets you offload work or get another model's opinion without losing visibility or control of the task.

Skill for Claude CodeCodex

Written for Claude Code and Codex: shipped in a Claude Code plugin, but also reads ~/.codex or $CODEX_HOME. Also seen: mentions subagents; mentions Claude Code; mentions Codex.

Part of the rageatc-code-oss plugin — 21 skills, 5 agents, 1 hook shipped together

Good fit Use it to delegate implementation, ask for a second opinion on design or reasoning, or run lighter and harder tasks with different Codex models.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/isvlasov/rageatc-oss/delegating-to-codex
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.

Any agent
npx skills add isvlasov/rageatc-oss --skill delegating-to-codex
Clone the repo
git clone --depth 1 https://github.com/isvlasov/rageatc-oss

Made for: Claude Code, Codex.

Or install rageatc-code-oss, the plugin that ships this one along with the rest of its 21 skills, 5 agents, 1 hook.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for delegating-to-codex

README.md
[![agentmods](https://agentmods.dev/badge/skills/isvlasov/rageatc-oss/delegating-to-codex/github.svg)](https://agentmods.dev/skills/isvlasov/rageatc-oss/delegating-to-codex)
Your own site
<a href="https://agentmods.dev/skills/isvlasov/rageatc-oss/delegating-to-codex"><img src="https://agentmods.dev/badge/skills/isvlasov/rageatc-oss/delegating-to-codex/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for delegating-to-codex

Your own site · 80×15
<a href="https://agentmods.dev/skills/isvlasov/rageatc-oss/delegating-to-codex"><img src="https://agentmods.dev/badge/skills/isvlasov/rageatc-oss/delegating-to-codex.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 103 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,020 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00103 $0.01020
Opus 5 $0.00051 $0.00510
Sonnet 5 $0.00021 $0.00204
Haiku 4.5 $0.00010 $0.00102

Measured 10d ago against content hash 30a5afd32bcd, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade B, and why

delegating-to-codex scanned grade B with 1 finding 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 10d ago.

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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

Pass both explicitly — the user's `~/.codex/config.toml` defaults may lag behind the current family. Current family (5.6, mid-2026 — verify with the user when in doubt):
plugins/rageatc-code-oss/skills/delegating-to-codex/SKILL.md · 55 lines

How it starts

The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Delegating to Codex

The subagent is an interactive Codex CLI session in a herdr pane, billed to the user's ChatGPT subscription. herdr provides the control loop — spawn, state, read, send — and its codex integration reports state automatically. The session stays visible and steerable; you own acceptance of its work.

Preflight

which codex && codex login status && herdr agent list >/dev/null && echo ok

Fixes: npm install -g @openai/codex to install; codex login if not logged in (interactive — hand to the user); herdr integration install codex if the integration is missing; herdr itself requires the user to be inside a herdr session.

Model and effort

Pass both explicitly — the user's ~/.codex/config.toml defaults may lag behind the current family. Current family (5.6, mid-2026 — verify with the user when in doubt):

Model Effort Use for
gpt-5.6-terra high Default worker — solid for most delegated tasks
gpt-5.6-sol medium Harder reasoning, design questions, second opinions
gpt-5.6-luna medium Quick, light tasks

Delegate

  1. Spawn with the task as the prompt argument — short kebab task name; cwd is the project the task touches. Pin the pane to your own workspace — spawning defaults to whichever workspace the user has focused at that moment. Do NOT send the initial task via herdr agent send: the starting TUI can swallow the head of the text (observed: a task beginning "Create a file zdravstvuy.txt" arrived as ".txt ..."). The prompt argument avoids this entirely:

    WS=$(herdr pane get "$HERDR_PANE_ID" | jq -r '.result.pane.workspace_id')
    herdr agent start <task-name> --cwd <dir> --workspace $WS --no-focus -- \
      codex -m <model-id> -c model_reasoning_effort="<effort>" --sandbox workspace-write -a never "<task>"
    

    Task text is single-line and self-contained: goal, relevant paths, done-criterion — Codex has not read your conversation. Capture pane_id from the JSON response. Add --search when the task needs the web. The sandbox/approval pair keeps the subagent autonomous but write-restricted to its workspace; escalate to a different sandbox mode only with the user's consent.

  2. Confirm it started: herdr agent wait <task-name> --status working --timeout 30000. On timeout, read the pane to see what state the session is in.

  3. Wait for completion: herdr agent wait <task-name> --status idle --timeout 600000 (scale to the task). The wait resolves on the next status change; read the status it reports: a finished run settles to idle, blocked means Codex is waiting on input (read the pane, respond via send + enter, wait again), and errors also settle — so always verify.

  4. Verify: herdr agent read <task-name> --lines 60 for the subagent's account, then check the actual outputs (files, command results) yourself. Acceptance is your judgement, not its claim.

  5. Steer if needed — by now the TUI is fully alive, so send is safe:

Read the full file on GitHub · 55 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. 10d ago First seen · 55 lines · 103 tokens per session scan B 30a5afd32bcd

Subscribe to this mod's changes

delegating-to-codex is a skill published in the GitHub repository isvlasov/rageatc-oss (9 stars, last pushed 1mo ago), licensed MIT. It adds 103 tokens to every session and 1,020 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

commit-optimization

Use when configuring settings.json to reduce fuse-commit-pro's context token usage.

fusengine/agents · 20 tokens

i-have-adhd

Shape output for a reader with ADHD: lead with the next action, number multi-step work, restate state across turns, suppress tangents, give specific time estimates, make wins visible. Invoke with /i-have-adhd; stays on until "stop adhd mode".

ayghri/i-have-adhd · 62 tokens

quick

Implement small Phoenix changes without planning — add validations, update routes, fix components, create migrations. Use for single-file edits under 50 lines.

oliver-kriska/claude-elixir-phoenix · 31 tokens

service-desk

Runs the IT service desk — intake, triage, prioritization, escalation, knowledge, and the metrics that improve service rather than distort it. Use this to set up or fix a service desk, design ticket priority and escalation, reduce repeat contacts, structure a knowledge base, or work out why a desk hitting its targets…

cbrock84/headcount · 73 tokens

collaboration-platform-administration

Administers the email, chat, meeting and file-sharing platform the organization runs on — tenant and domain configuration, mail authentication and routing, phishing and spam controls, shared mailboxes and distribution groups, external sharing and guest access, permission sprawl in file storage, and retention and legal…

cbrock84/headcount · 93 tokens

session-cleanup

Identify stale and empty Claude Code sessions in the Agent Monitor and explain the cleanup endpoint (POST /api/settings/cleanup), always showing the exact list of what WOULD be removed before anything is deleted. Cleanup permanently deletes data, so this skill previews first and requires explicit user confirmation.…

hoangsonww/Claude-Code-Agent-Monitor · 67 tokens