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 rsalmn/ExtremeRouter --skill extremerouter-sttgit clone --depth 1 https://github.com/rsalmn/ExtremeRouterWrote 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/rsalmn/extremerouter/extremerouter-stt)<a href="https://agentmods.dev/skills/rsalmn/extremerouter/extremerouter-stt"><img src="https://agentmods.dev/badge/skills/rsalmn/extremerouter/extremerouter-stt.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
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 →
- medium Agent Snooping · line 8 Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.Fix: Remove all code or instructions that list or read other skills' files or directories. Skills should operate independently; cross-skill access is a privilege escalation.
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 | $0.00064 | $0.00968 |
| Opus 5 | $0.00032 | $0.00484 |
| Sonnet 5 | $0.00013 | $0.00194 |
| Haiku 4.5 | $0.00006 | $0.00097 |
Grade A, and why
extremerouter-stt scanned grade A 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 8d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl $NINEROUTER_URL/v1/models/stt | jq '.data[].id' How it starts
The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ExtremeRouter — Speech-to-Text
Requires NINEROUTER_URL (and NINEROUTER_KEY if auth enabled). See https://raw.githubusercontent.com/rsalmn/extremerouter/refs/heads/master/skills/extremerouter/SKILL.md for setup.
Discover
curl $NINEROUTER_URL/v1/models/stt | jq '.data[].id'
# Per-model params (language, response_format, prompt, temperature support)
curl "$NINEROUTER_URL/v1/models/info?id=openai/whisper-1"
model = STT model ID (e.g. openai/whisper-1, groq/whisper-large-v3, deepgram/nova-3, gemini/gemini-2.5-flash).
Endpoint
POST $NINEROUTER_URL/v1/audio/transcriptions (OpenAI Whisper compatible, multipart/form-data)
| Field | Required | Notes |
|---|---|---|
model |
yes | from /v1/models/stt |
file |
yes | audio file (mp3, wav, m4a, webm, ogg, flac) |
language |
no | ISO-639-1 (e.g. en, vi) |
prompt |
no | hint text to guide transcription |
response_format |
no | json (default) / text / verbose_json / srt / vtt |
temperature |
no | 0–1 |
Examples
curl -X POST "$NINEROUTER_URL/v1/audio/transcriptions" \
-H "Authorization: Bearer $NINEROUTER_KEY" \
-F "model=openai/whisper-1" \
-F "[email protected]" \
-F "language=vi"
JS (Node):
import { createReadStream } from "node:fs";
const form = new FormData();
form.append("model", "groq/whisper-large-v3-turbo");
form.append("file", new Blob([await (await import("node:fs/promises")).readFile("audio.mp3")]), "audio.mp3");
const r = await fetch(`${process.env.NINEROUTER_URL}/v1/audio/transcriptions`, {
method: "POST",
headers: { "Authorization": `Bearer ${process.env.NINEROUTER_KEY}` },
body: form,
});
const { text } = await r.json();
console.log(text);
Response shape
Default (response_format=json):
{ "text": "Xin chào, đây là bản ghi âm." }
verbose_json adds language, duration, segments[] with timestamps.
srt / vtt return subtitle text.
Provider quirks
| Provider | model format |
Notes |
|---|---|---|
openai |
whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe |
Native OpenAI shape |
groq |
whisper-large-v3, whisper-large-v3-turbo, distil-whisper-large-v3-en |
Fastest; OpenAI shape |
gemini |
gemini-2.5-flash, gemini-2.5-pro, gemini-2.5-flash-lite |
Server converts to generateContent with audio inline |
deepgram |
nova-3, nova-2, whisper-large |
Token auth; server adapts response |
assemblyai |
universal-3-pro, universal-2 |
Async upload+poll handled server-side |
nvidia |
nvidia/parakeet-ctc-1.1b-asr |
NIM endpoint |
huggingface |
openai/whisper-large-v3, openai/whisper-small |
HF Inference API |
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.
- 8d ago First seen · 80 lines · 64 tokens per session scan A 5e204fac2745
extremerouter-stt is a skill published in the GitHub repository rsalmn/ExtremeRouter (31 stars, last pushed yesterday), licensed MIT. It adds 64 tokens to every session and 968 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
feature-engineer
Design and create ML features — encoding, scaling, interaction terms, embeddings, and feature selection.
gcp-architect
Design Google Cloud architectures — GKE, Cloud Run, BigQuery, Pub/Sub, and Cloud Spanner solutions.
midjourney-prompter
Engineer Midjourney prompts — style references, aspect ratios, negative prompts, and v6 parameter tuning.
ml-pipeline
Build end-to-end ML pipelines — data prep, feature engineering, model training, evaluation, and MLflow tracking.
model-evaluator
Evaluate ML models rigorously — cross-validation, confusion matrices, ROC curves, bias audits, and interpretability.
pandas-helper
Master pandas operations — DataFrames, Series, groupby, merge, time-series resampling, and memory optimization.