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.
npx agentmods add skills/youngjaedev/my-claude-plugins/codex-imagenpx skills add YoungjaeDev/my-claude-plugins --skill codex-imagegit clone --depth 1 https://github.com/YoungjaeDev/my-claude-pluginsWrote 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.
[](https://agentmods.dev/skills/youngjaedev/my-claude-plugins/codex-image)<a href="https://agentmods.dev/skills/youngjaedev/my-claude-plugins/codex-image"><img src="https://agentmods.dev/badge/skills/youngjaedev/my-claude-plugins/codex-image.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00098 | $0.03263 |
| Opus 5 | $0.00049 | $0.01631 |
| Sonnet 5 | $0.00020 | $0.00653 |
| Haiku 4.5 | $0.00010 | $0.00326 |
Grade A, and why
codex-image 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 4d 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.
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.
How it starts
The opening of the file, as written. The whole thing — 175 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Codex Image
Generate or edit raster images by asking Codex CLI to use its image generation capability. This skill is a Claude Code bridge: do not call the OpenAI REST API, do not ask for an API key, and do not clone or run third-party skill code.
Defaults
- Generation needs explicit grounding: a direct user request, or a task spec that names codex-image as the image path. Without that grounding — or when scope/count is ambiguous — ask before generating, because image generation has cost and side effects.
- Default output directory:
assets/generated/codex-image/under the project root. - Default size:
auto. - Default quality:
auto. - Default count:
1. - Keep generated filenames unique and non-destructive:
codex-image-YYYYMMDD-HHMMSS[-N].png. - Treat
-nas variations of one prompt. For distinct assets, run separate generations.
Argument Handling
Invocation arguments:
$ARGUMENTS
Parse the invocation arguments manually:
--size:autoorWIDTHxHEIGHT.--quality:low,medium,high, orauto.--out: output directory. Resolve relative paths from the project root.-n: number of variants. Use1through4without extra confirmation; ask before generating more.--edit: local image path to use as the base to edit — the prompt describes the change. Like every other run, the output is still saved as a new, non-destructive file in the output directory (see Defaults / Result Handling); the input at--edit's path is never overwritten. Verify it exists and read it before delegating.--ref: local image path to attach as a style/character reference while generating an image whose subject/scene is otherwise new (not a modification of the reference's own scene) — distinct intent from--edit, though the output is a new file either way. Verify it exists and read it before delegating.--editand--refare mutually exclusive (editing a specific image vs. using one as inspiration for a different image are different intents) — if both are given, ask which one is meant. Unverified mechanism:codex exec's-i, --image <FILE>...flag (confirmed viacodex exec --helpon codex-cli 0.142.3) is a generic "attach image(s) to the prompt" transport — it does not itself distinguish edit-in-place from reference-only. The edit-vs-reference distinction relies entirely on how the prompt text frames the intent (see Prompt Handoff), not on a separate CLI mechanism. This has not been empirically verified with a live generation in this repo; on first real use, check the result actually looks like a new image guided by the reference (not a near-copy or a literal edit of it) and report back if it doesn't behave as expected.--model: Codex model id, passed through ascodex exec -m <id>. Omit to use Codex's own default model — that auto-tracks the latest model, so no model pin is maintained here.--reasoning: reasoning effort, passed through as-c model_reasoning_effort="<effort>"(e.g.low,medium,high,xhigh). Explicit opt-in only — image generation barely benefits from reasoning effort, so omit unless the user asks for it.--sandbox: Codex sandbox mode for the run:read-only,workspace-write(default), ordanger-full-access. The default staysworkspace-write(enough to save the PNG); escalate only on explicit request. The full approval+sandbox bypass is--dangerously-bypass-approvals-and-sandbox— pass it only when the user explicitly asks, never by default.- Remaining text is the image prompt.
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.
- 4d ago First seen · 175 lines · 98 tokens per session scan A 58a5c0aae4bb
codex-image is a skill published in the GitHub repository YoungjaeDev/my-claude-plugins (2 stars, last pushed 9d ago), licensed MIT. It adds 98 tokens to every session and 3,263 once invoked, about $0.0005 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.
Other skills, from other repositories
general-video
Author or edit a custom HyperFrames composition when no specialized workflow fits, or when BRIEF.md sets flow: companion. Use for longer or multi-scene pieces, brand and sizzle reels, montages, static loops, static title cards, footage remixes, and freeform builds. Use motion-graphics instead for a short unnarrated…
diagnostic-stem-delivery
Audio production with diagnostic analysis, timecode parsing from documents, and verified export workflow.
chengfeng-cut
剪辑中文口播原素材：逐词转录、词典修字出修字表、五轮扫描找口误与重复、汇总表与重复句子表、打开 Studio 让用户复核、复盘沉淀用户偏好与词典。只产出一份已复核的删词账本，不切媒体、不做字幕、不做分镜动画。用户说剪口播、处理口误、生成口播基础素材、继续剪口播，或确认卡回传 action=returncutreview 时使用。不要用于执行物理剪切、导出剪后视频、单独安装、单独打开工作台或口播分镜成片。.
chengfeng-check-updates
剪辑环境的唯一管理者：就绪检查（skills 是否最新 → Runtime 是否配套）、Skills 更新激活、Runtime 安装与体检。用户说检查更新、安装剪辑环境、装播放器、检查剪辑环境、剪辑环境就绪了吗、配置转录凭证时使用；业务 Skill（剪口播/字幕/画面/导出）第 0 步也引用本 Skill 的就绪检查。不用于剪辑、字幕、画面、导出本身或项目数据迁移。.
infographic-template-updater
Update template catalogs and UI prompts after adding new infographic templates (src/templates/.ts), including SKILL.md template list, site gallery template mappings, and the AIPlayground prompt list.
vox-director
Turn ONE topic into a finished Vox-style paper-collage explainer / ad video, end to end on the Atlas Cloud API + local ffmpeg — script, collage keyframes, motion, voice-over, music, captions, all automated. Use this whenever the user wants a "Vox style" video, a paper/torn-paper collage animation, a "motion collage"…