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 arjunkr303/multi-llm-mcp --skill multi-llm-skillgit clone --depth 1 https://github.com/arjunkr303/multi-llm-mcpWrote 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/arjunkr303/multi-llm-mcp/multi-llm-skill)<a href="https://agentmods.dev/skills/arjunkr303/multi-llm-mcp/multi-llm-skill"><img src="https://agentmods.dev/badge/skills/arjunkr303/multi-llm-mcp/multi-llm-skill/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/arjunkr303/multi-llm-mcp/multi-llm-skill"><img src="https://agentmods.dev/badge/skills/arjunkr303/multi-llm-mcp/multi-llm-skill.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00118 | $0.01186 |
| Opus 5.5 | $0.00047 | $0.00474 |
| Sonnet 5.5 | $0.00024 | $0.00237 |
| Haiku 4.5 | $0.00012 | $0.00119 |
Grade A, and why
multi-llm 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 7d 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 https://integrate.api.nvidia.com/v1/models -H "Authorization: Bearer $NVIDIA_API_KEY" How it starts
The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Multi-LLM delegation
Delegate a subtask to an external LLM by running the bundled script. It works with only Python installed — no pip installs, no server process, no MCP setup.
When to delegate (and when not to)
Delegate when a second independent perspective adds value: reviewing your own work, stress-testing an argument, generating a variant implementation, or getting specialized reasoning (e.g. a large-context model over many files). Do not delegate trivial lookups you can answer directly, and do not delegate work that requires reading this user's filesystem — paste the relevant code into the prompt instead, because the external model cannot see any files.
How to call it
python "<skill-dir>/scripts/ask_llm.py" --provider <p> --model <m> --system-prompt "<persona>" --prompt "<task + code>"
Pin <skill-dir> to the actual directory containing this SKILL.md. For long
code, write it to a temp file and pass --prompt-file <path> (or pipe: use
--prompt-file -). Read the script's stdout as the model's answer; stderr
carries errors. Exit codes: 0 success, 1 misconfiguration (e.g. missing API
key), 2 CLI usage error or network/API failure.
Provider routing
| Situation | Provider | Model | Temp |
|---|---|---|---|
| Fast, cheap review or quick question | groq |
llama-3.1-8b-instant |
0.1 |
| Test generation, careful code edits | deepseek |
deepseek-coder |
0.0 |
| Many files / large context, strongest reasoning | openrouter |
anthropic/claude-3.5-sonnet |
0.0 |
| Proprietary code that must not leave the machine | ollama |
codellama (any pulled model) |
0.0 |
| Big open-weights model | nvidia |
meta/llama-3.2-11b-vision-instruct |
0.0 |
Default when the user has no preference: check --list-providers first, and
prefer the fastest provider whose key is actually present — typically groq,
else nvidia, else ollama (no key needed).
Model names drift — older ones get retired (NVIDIA marks dead models with
HTTP 410) and providers publish new ones. Note that NVIDIA's /v1/models
catalog also lists models the calling account may not actually be enabled
for: a 404 "Function ... Not found for account" means that model is not
available to you — pick another from the catalog. Query the live catalog,
then fall back on errors:
What ships with it
2 files 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.
- 7d ago First seen · 92 lines · 118 tokens per session scan A 55b1e4fce3c5
multi-llm is a skill published in the GitHub repository arjunkr303/multi-llm-mcp (3 stars, last pushed 8d ago), licensed MIT. It adds 118 tokens to every session and 1,186 once invoked, about $0.0005 per session on Opus 5.5. 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 2026-10-21.
Other skills, from other repositories
test-review
You are an expert DataHub test reviewer. Your role is to evaluate pytest smoke tests against established testing standards, identify issues, and provide actionable feedback.
grade-tests
Grade curated individual tests for readiness, A-F quality and improvements. ALWAYS USE FOR: grade tests, review only a named test, per-test readiness or quality bands for supplied methods, bodies, file spans or bounded PR diffs, including existing tests. Return Pass/Failed/Uncertain/Not applicable; unresolved/empty…
go-testing
Trigger: Go tests, go test coverage, Bubbletea teatest, golden files. Apply focused Go testing patterns.
brooks-test
Test quality review drawing on twelve classic engineering books — with primary focus on xUnit Test Patterns, The Art of Unit Testing, How Google Tests Software, and Working Effectively with Legacy Code — that diagnoses structural problems in an existing test suite: brittleness, mock abuse, coverage illusions, slow…
dart-mechanical-refactor
DART Mechanical Refactor: perform a behavior-preserving mechanical refactor.
refactor
Refactors code for quality and maintainability. Triggers: refactor, clean up, restructure, improve code, modernize.