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 skills add PracticalSwan/agent-skills --skill codex-router-mediagit clone --depth 1 https://github.com/PracticalSwan/agent-skillsWrote 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/practicalswan/agent-skills/codex-router-media)<a href="https://agentmods.dev/skills/practicalswan/agent-skills/codex-router-media"><img src="https://agentmods.dev/badge/skills/practicalswan/agent-skills/codex-router-media/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.
<a href="https://agentmods.dev/skills/practicalswan/agent-skills/codex-router-media"><img src="https://agentmods.dev/badge/skills/practicalswan/agent-skills/codex-router-media.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Prompt Injection · line 61 Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.Fix: Audit all comments and invisible characters. Remove any instructions that direct the agent to perform unauthorized actions. Use plain, reviewable content.
- medium Rogue Agent · line 6 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
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 5.1 | $0.00097 | $0.01357 |
| Opus 5 | $0.00048 | $0.00678 |
| Sonnet 5 | $0.00019 | $0.00271 |
| Haiku 4.5 | $0.00010 | $0.00136 |
Grade A, and why
codex-router-media 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.
How it starts
The opening of the file, as written. The whole thing — 109 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MiniMax media generation (codex-router)
The router stores the operator's MiniMax Token Plan API key. The media
command resolves that key itself — you never see or handle the credential.
Each call spends the operator's paid MiniMax quota, so generate only what the
user explicitly asked for, once, and reuse the downloaded file for retries of
later steps.
How to call it
Run the router CLI through the shell tool. Locate it once per session: the
install root is the current.sourceRoot field of
~/.codex/codex-router/install-manifest.json (%USERPROFILE% on Windows),
and the command is <sourceRoot>/bin/media (<sourceRoot>\model-router.ps1 codex media on Windows). Quote the path; it may contain spaces. If the
manifest is missing, try ~/.local/share/codex-router/bin/media.
Always pass --json so the result is machine-readable, and --out so the
file lands where the user wants it (default: the current directory).
Actions
media video --prompt "TEXT" [--duration 6] [--resolution 768P|1080P] [--image PATH_OR_URL] [--out clip.mp4] --json
media music --prompt "style, mood, scenario" (--lyrics "[verse]..." | --instrumental) [--out track.mp3] --json
media speech --text "TEXT" [--voice male-qn-qingse] [--speed 1.0] [--out voice.mp3] --json
media image --prompt "TEXT" [--ratio 16:9] [--count 1] [--out picture.jpeg] --json
media status --task-id ID --json
- video is asynchronous upstream: the command submits a task and polls
until the clip is rendered (typically 1–3 minutes; it blocks until done and
downloads the mp4). If it exits saying the task is still rendering, wait and
run
media status --task-id ID— never resubmit the same prompt. - video model: the default
MiniMax-Hailuo-02is the one the Token Plan serves.MiniMax-H3is refused on this subscription; the command already explains that, so do not retry with H3. - music needs either
--lyrics(use section tags like[verse],[chorus]) or--instrumental. --imageturns video generation into image-to-video from a first frame (local file path or URL).
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- yesterday Changed 5bfdd0f67f8e
- 3d ago Changed 0ecce5d6ae2a
- 6d ago Changed 6c03252f3ba7
- 9d ago First seen · 109 lines · 97 tokens per session scan A 969efcbcdf1c
codex-router-media is a skill published in the GitHub repository PracticalSwan/agent-skills (13 stars, last pushed yesterday), licensed MIT. It adds 97 tokens to every session and 1,357 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 2026-08-30.
Other skills, from other repositories
moq
Build live video, audio, and real-time data apps with Media over QUIC (MoQ). Use when adding live streaming, conferencing, voice AI, or real-time pub/sub to an app; when integrating the @moq/ npm packages, moq- Rust crates, or the Python/Kotlin/Swift/Go/C bindings; or when running a moq-relay server or a gateway…
tiktok-transcribe
A workflow that downloads TikTok videos, turns their audio into text with speech-recognition software, and saves the result as a Markdown file.
atlas-cloud-media
Generate Atlas Cloud images and videos through its asynchronous media API with schema-first model selection and credential-safe polling.
bilibili-transcribe
A workflow that downloads videos from Bilibili, a Chinese video-sharing site, transcribes their speech, and saves the result as Markdown text. It accepts Bilibili links or BV video identifiers and tries available subtitles before audio transcription.
douyin-transcribe
A workflow for downloading Douyin videos, converting their speech to text with SenseVoice-Small, and saving the result as a Markdown file. Douyin is the Chinese short-video platform also known internationally as TikTok’s Chinese counterpart.
media-expert
Expert-level media production, content management, streaming, broadcasting, and media technology systems. Use when the user mentions video, streaming, broadcast, CMS, or production, or when the task involves Media Production, Streaming and Broadcasting, Technologies, or Standards and Protocols.