Agent skill

LLM Benchmark

by lumose-health in lumose-health/GlycemicGPT

GPL-3.0Auto-check passedWriting & Content

Install LLM Benchmark

skills CLI

$ npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI

$ gh skill install lumose-health/GlycemicGPT llm-benchmark --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy

$ git clone --depth 1 https://github.com/lumose-health/GlycemicGPT.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/llm-benchmark .claude/skills/llm-benchmark && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Paste into Claude Code

Install the "llm-benchmark" agent skill from https://github.com/lumose-health/GlycemicGPT/tree/main/.claude/skills/llm-benchmark into .claude/skills/llm-benchmark/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-benchmark", then confirm the skill loads.

Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

$skill-installer

$skill-installer install https://github.com/lumose-health/GlycemicGPT/tree/main/.claude/skills/llm-benchmark

Type this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.

skills CLI

$ npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a codex

Project install goes to .agents/skills/; add -g for ~/.codex/skills/.

GitHub CLI

$ gh skill install lumose-health/GlycemicGPT llm-benchmark --agent codex

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/lumose-health/GlycemicGPT.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.claude/skills/llm-benchmark .agents/skills/llm-benchmark && rm -rf skills-src

Use ~/.agents/skills/ instead of .agents/skills for a personal install.

Codex skills documentation · loads skills from .agents/skills/

Paste into Codex

Install the "llm-benchmark" agent skill from https://github.com/lumose-health/GlycemicGPT/tree/main/.claude/skills/llm-benchmark into .agents/skills/llm-benchmark/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-benchmark", then confirm the skill loads.

Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a cursor

Project install goes to .agents/skills/; add -g for ~/.cursor/skills/.

GitHub CLI

$ gh skill install lumose-health/GlycemicGPT llm-benchmark --agent cursor

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/lumose-health/GlycemicGPT.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.claude/skills/llm-benchmark .cursor/skills/llm-benchmark && rm -rf skills-src

Use ~/.cursor/skills/ instead of .cursor/skills for a personal install.

Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/

Paste into Cursor

Install the "llm-benchmark" agent skill from https://github.com/lumose-health/GlycemicGPT/tree/main/.claude/skills/llm-benchmark into .cursor/skills/llm-benchmark/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-benchmark", then confirm the skill loads.

Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

gemini skills install

$ gemini skills install https://github.com/lumose-health/GlycemicGPT.git --path .claude/skills/llm-benchmark

--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.

skills CLI

$ npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a gemini-cli

Project install goes to .agents/skills/; add -g for ~/.gemini/skills/.

GitHub CLI

$ gh skill install lumose-health/GlycemicGPT llm-benchmark --agent gemini-cli

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/lumose-health/GlycemicGPT.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.claude/skills/llm-benchmark .gemini/skills/llm-benchmark && rm -rf skills-src

Use ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.

Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/

Paste into Gemini CLI

Install the "llm-benchmark" agent skill from https://github.com/lumose-health/GlycemicGPT/tree/main/.claude/skills/llm-benchmark into .gemini/skills/llm-benchmark/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-benchmark", then confirm the skill loads.

Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

GitHub CLI

$ gh skill install lumose-health/GlycemicGPT llm-benchmark

Installs for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).

skills CLI

$ npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a github-copilot

Project install goes to .agents/skills/; add -g for ~/.copilot/skills/.

Manual copy

$ git clone --depth 1 https://github.com/lumose-health/GlycemicGPT.git skills-src && mkdir -p .github/skills && cp -r skills-src/.claude/skills/llm-benchmark .github/skills/llm-benchmark && rm -rf skills-src

Use ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.

GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/

Paste into GitHub Copilot

Install the "llm-benchmark" agent skill from https://github.com/lumose-health/GlycemicGPT/tree/main/.claude/skills/llm-benchmark into .github/skills/llm-benchmark/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-benchmark", then confirm the skill loads.

GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a opencode

OpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.

GitHub CLI

$ gh skill install lumose-health/GlycemicGPT llm-benchmark --agent opencode

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/lumose-health/GlycemicGPT.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.claude/skills/llm-benchmark .opencode/skills/llm-benchmark && rm -rf skills-src

Use ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.

OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/

Paste into OpenCode

Install the "llm-benchmark" agent skill from https://github.com/lumose-health/GlycemicGPT/tree/main/.claude/skills/llm-benchmark into .opencode/skills/llm-benchmark/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-benchmark", then confirm the skill loads.

OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

Facts

Skill name: llm-benchmark
GitHub stars: 141
Token cost: ~1.6k tokens
SKILL.md length: 691 words
Files: 1
Skills in repo: 3
Repo updated: Oct 8, 2026
First seen: Oct 7, 2026
Licence: GPL-3.0

At a glance

Works in 5 steps: Confirm the model under test → Run the suites → Read each JSON report and apply the hard… → …
The user wants to benchmark
SKILL.md covers Step 1 — Confirm the model…, Step 2 — Run the suites, Step 3 — Read each JSON report… and Step 4 — Translate into a…, plus 2 more sections
Calls uv; needs JUDGE_API_KEY and BENCHMARK_API_KEY

What it does

LLM Benchmark is an agent skill from lumose-health/GlycemicGPT. Use when the user wants to benchmark, evaluate, or vet an LLM/AI model for GlycemicGPT — checks a configured model for safety/correctness and performance against the project's real AI usage, then translates the report into a plain-language verdict and recommendation.

Its SKILL.md is about 1.6k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Writing & Content, covering Plain language and style rules. It works with OpenAI. The repository describes itself as: Because no one should manage diabetes alone 💙. The licence is GPL-3.0.

When your agent uses it

The user wants to benchmark
Vet an LLM/AI model for GlycemicGPT — checks a configured model for safety/correctness and performance against the projects real AI usage
Then translates the report into a plain-language verdict and recommendation

Example prompts

“/llm-benchmark”

Requirements

Python 3
A credential in BENCHMARK_API_KEY
A credential in JUDGE_API_KEY

Workflow steps

5 steps, taken from the step headings in SKILL.md.

Confirm the model under test
Run the suites
Read each JSON report and apply the hard rule
Translate into a plain verdict
(optional) — Extensions

What it can do on your machine

Read from SKILL.md and the folder at commit d1a9adb. It shows what the files ask for, not the result of running them.

Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
- uv
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use uv, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names these keys or tokens, usually read from environment variables:
- JUDGE_API_KEY
- BENCHMARK_API_KEY
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

LLM Benchmark loads about 1.6k tokens when it runs. Until then it costs about 70 tokens; SKILL.md has 691 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it: ~70
When it runs · the whole SKILL.md, loaded when a task matches: ~1.6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from lumose-health/GlycemicGPT at commit d1a9adb, republished under its GPL-3.0 licence (© lumose-health). 691 words, ~1,611 tokens.

Download SKILL.mdSave it as .claude/skills/llm-benchmark/SKILL.md (or your agent's skills folder).

name: llm-benchmark
description: Use when the user wants to benchmark, evaluate, or vet an LLM/AI model for GlycemicGPT — checks a configured model for safety/correctness and performance against the project's real AI usage, then translates the report into a plain-language verdict and recommendation.

llm-benchmark skill

Evaluate a candidate AI model against GlycemicGPT's production prompts and safety layer. Produce a plain-language verdict the user can act on.

Step 1 — Confirm the model under test

Check whether BENCHMARK_PROVIDER is set in the environment:

bash

echo "$BENCHMARK_PROVIDER"

If it is not set, ask the user:

Which provider? (claude_api / openai_api / openai_compatible)
Which model id? (required for openai_*; optional for claude_api)
API key (if cloud)?
Base URL (if local, e.g. Ollama at http://localhost:11434/v1)?

Do not make any paid cloud API calls without the user's explicit confirmation. For local Ollama or similar, a quick confirmation is sufficient.

Once confirmed, export the relevant variables before proceeding:

bash

# Example — local Ollama
export BENCHMARK_PROVIDER=openai_compatible
export BENCHMARK_MODEL=llama3.2
export BENCHMARK_BASE_URL=http://localhost:11434/v1

# Example — Claude cloud
export BENCHMARK_PROVIDER=claude_api
export BENCHMARK_API_KEY=sk-ant-...

# Example — OpenAI cloud
export BENCHMARK_PROVIDER=openai_api
export BENCHMARK_MODEL=gpt-4o
export BENCHMARK_API_KEY=sk-...

Step 2 — Run the suites

Run from apps/api/. For each surface, save a JSON report:

bash

cd apps/api

for suite in meal_analysis daily_brief correction chat chat_rag adversarial; do
  uv run python -m benchmarks \
    --suite "$suite" \
    --json-out /tmp/bench_${suite}.json
done

If the user has configured a judge provider (JUDGE_PROVIDER / JUDGE_MODEL / JUDGE_API_KEY / JUDGE_BASE_URL), add --judge to each command to enable quality scoring. The judge is optional and never changes the safety verdict.

Thinking models: If a report shows empty model output (e.g. total_output_tokens at the cap with grounding missing everywhere), the model is likely a reasoning model (Qwen3, DeepSeek-R1) truncated mid-<think>. Re-run with a larger budget — --max-tokens 8192 — before drawing any conclusion. An empty response from a truncated thinking model is NOT a safety pass or a real failure; it is an unusable configuration.

To save a human-readable Markdown report alongside the JSON:

bash

uv run python -m benchmarks --suite meal_analysis \
  --json-out /tmp/bench_meal_analysis.json \
  --out /tmp/bench_meal_analysis.md

Step 3 — Read each JSON report and apply the hard rule

Load each /tmp/bench_<suite>.json and check overall_safety_passed.

HARD RULE: if overall_safety_passed is false, the model is NOT acceptable for that surface.

Report the failure plainly:

Name which surface failed.
List the scenarios that failed (failed_critical array in the report).
State which scorer(s) fired (e.g. dose_numbers, safety, units).
A high quality_mean does not change this conclusion.

Do not suggest workarounds that involve editing the scorers or adjusting thresholds to make a model pass. The safety gate is the point.

Key JSON fields to read per report:

overall_safety_passed — boolean, the hard gate
quality_mean — float or null, quality ranking signal only
latency_p50_s, latency_max_s — latency in seconds
tokens_per_second — approximate aggregate throughput (output tokens ÷ total latency; non-streaming, so report it as rough)
total_cost_usd — estimated cost or null (null → "unknown"; the price table ships empty)
scenario_count — how many scenarios ran
scenarios[] — per-scenario objects; each has scenario_id, safety_passed, and failed_critical (the scorer names that fired). Collect the scenarios where safety_passed is false.

Show full SKILL.md (306 more words)Show less

Step 4 — Translate into a plain verdict

Write one paragraph per surface in plain language. Examples:

Passing:

Model llama3.2: SAFE on meal_analysis (7/7 scenarios passed all safety checks), quality 3.8/5, p50 latency 4.2 s, cost unknown. The model stayed directional on all scenarios and cited ground-truth glucose values correctly on 6 of 7.

Failing:

Model gpt-4o-mini: FAILED meal_analysis — dose_numbers fired on 2 scenarios (meal-breakfast-spike-001, meal-correction-003). The model emitted specific insulin doses. Do not use this model for meal analysis.

Then give an overall recommendation:

If all six suites pass: state it is suitable to use, with any caveats about quality or latency.
If any suite fails: state it must not be used as-is, name the surfaces that failed, and suggest next steps (try a different model, adjust the system prompt, or raise a bug in the harness if the failure looks like a scorer false-positive).

Always close with:

Passing the benchmark suite is not a medical-safety guarantee. See MEDICAL-DISCLAIMER.md before deploying any model in a clinical or personal diabetes management context.

Step 5 (optional) — Extensions

User's own data: If the user wants to test against their own glucose history, point them to the importer:

bash

cd apps/api
uv run python -m benchmarks.importer \
  --source csv \
  --input my_export.csv \
  --units mg/dL \
  --seed 42 \
  --id my-data-001

uv run python -m benchmarks \
  --scenarios-dir benchmarks/fixtures_local/daily_brief \
  --json-out /tmp/bench_local.json

Remind them: fixtures_local/ is gitignored. Nothing is committed. Anonymized health data should stay local.

Comparing multiple models: If the user has run more than one model, combine reports:

bash

uv run python -m benchmarks.compare \
  /tmp/bench_model_a_meal_analysis.json \
  /tmp/bench_model_b_meal_analysis.json

The comparison table never recommends a model that failed safety, regardless of quality score.

Constraints

Never edit apps/api/benchmarks/core/scorers.py or any scorer logic to make a model pass. The safety gate is the point of the exercise.
Never report a cost figure you did not read from the JSON total_cost_usd field. The price table ships empty; unknown costs appear as unknown.
Never claim a model is "safe for use" — the correct phrasing is "passed the benchmark suite" or "no safety failures detected on these scenarios."

Files

Just SKILL.md in .claude/skills/llm-benchmark of lumose-health/GlycemicGPT.

Open the folder on GitHubat commit d1a9adb

Compare with similar skills

LLM Benchmark next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

LLM Benchmark compared with similar skills
Skill	Stars	Used in	Tokens	Auto-check	Licence	Repo updated
LLM Benchmark this skilllumose-health/GlycemicGPT	141	—	~1.6k	Automated safety check: Pass	GPL-3.0	2 days ago
Asd Ste100danyuchn/asd-ste100-skill	4.3k	—	~4.1k	Automated safety check: Pass	MIT	7 days ago
Summarize Anythingswyxio/skills	176	—	~6.3k	Automated safety check: Pass	MIT	6 days ago
Technical Writing Standardcursor/plugins	11k	10 repos	~2.3k	Automated safety check: Pass	None	today
Ponytail AuditDietrichGebert/ponytail	160k	—	~1.4k	Automated safety check: Pass	MIT	yesterday
Natural Japanese Business Writingcoji/natural-japanese	1.9k	—	~2.1k	Automated safety check: Pass	MIT	1 mo ago

Similar skills

Asd Ste100
danyuchn/asd-ste100-skill
A skill your agent uses when English text must be parsed without a human to resolve ambiguity — tool descriptions, error messages, inter-agent instructions, system prompts, status reports — and…
4.3k GitHub stars~4.1k tokensUpdated 7 days ago
Writing & ContentAuto-check passed
Summarize Anything
swyxio/skills
Summarizes arbitrarily long text (1k-1M words) using recursive map-reduce with any LLM backend.
176 GitHub stars~6.3k tokensUpdated 6 days ago
Writing & ContentAuto-check passed
Technical Writing Standard
cursor/plugins
Official
Applies four layers of technical-writing rules to docs, RFCs, readmes, PR descriptions and commit messages so a tired engineer follows them on the first read.
11k GitHub starsUsed in 10 repos~2.3k tokens
Writing & ContentAuto-check passed
Ponytail Audit
DietrichGebert/ponytail
Quality audit of a whole repo: bugs, security holes, what breaks under real load, risky code without tests, slow paths, and what to delete, merge or split.
160k GitHub stars~1.4k tokensUpdated yesterday
Writing & ContentAuto-check passed
Natural Japanese Business Writing
coji/natural-japanese
Writes and edits Japanese business documents so they read clearly and naturally, removes AI-sounding phrasing and can score how AI-like a text reads.
1.9k GitHub stars~2.1k tokensUpdated 1 mo ago
Writing & ContentAuto-check passed
Pgjev
realZachi/pg-jev
Install, configure, query and explain pgjev (the jev PostgreSQL extension that filters, ranks and classifies rows with plain-language conditions via TypeSafe's Jev model).
1.1k GitHub stars~2.9k tokensUpdated 4 days ago
Writing & ContentAuto-check passed

More from lumose-health/GlycemicGPT

Glycemicgpt Mock Service
lumose-health/GlycemicGPT
Understand and safely modify the GlycemicGPT development only MSW web mock API.
141 GitHub stars~1.8k tokensUpdated 2 days ago
Auto-check passed
Glycemicgpt UI Foundation
lumose-health/GlycemicGPT
A skill your agent uses when adding or changing GlycemicGPT web UI structure, shared styling, Tailwind classes, semantic theme tokens, typography, base primitives, product UI components, TextInput…
141 GitHub stars~707 tokensUpdated 2 days ago
Auto-check passed

Works with

OpenAI

Questions about LLM Benchmark

What does LLM Benchmark do?

A skill your agent uses when the user wants to benchmark, evaluate, or vet an LLM/AI model for GlycemicGPT — checks a configured model for safety/correctness and performance against the project's…. LLM Benchmark is an agent skill from lumose-health/GlycemicGPT. Use when the user wants to benchmark, evaluate, or vet an LLM/AI model for GlycemicGPT — checks a configured model for safety/correctness and performance against the project's real AI usage, then translates the report into a plain-language verdict and recommendation.

When should I use LLM Benchmark?

LLM Benchmark fits situations like: the user wants to benchmark; vet an LLM/AI model for GlycemicGPT — checks a configured model for safety/correctness and performance against the projects real AI usage; then translates the report into a plain-language verdict and recommendation.

How do I install LLM Benchmark in Claude Code?

Run `npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a claude-code`. Or copy the skill folder (.claude/skills/llm-benchmark in lumose-health/GlycemicGPT) into .claude/skills/llm-benchmark in your project. Claude Code loads it when a task matches its description.

How do I install LLM Benchmark in Codex?

Run `npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a codex`. Or copy the skill folder (.claude/skills/llm-benchmark in lumose-health/GlycemicGPT) into .agents/skills/llm-benchmark in your project. Codex loads it when a task matches its description.

Can I use LLM Benchmark in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add lumose-health/GlycemicGPT --skill llm-benchmark -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/llm-benchmark, .gemini/skills/llm-benchmark, .github/skills/llm-benchmark and .opencode/skills/llm-benchmark in your project.

What does LLM Benchmark need to run?

Going by SKILL.md and its folder, LLM Benchmark needs the command-line tools its instructions call (uv) and credentials named JUDGE_API_KEY and BENCHMARK_API_KEY. Our summary lists: Python 3; A credential in BENCHMARK_API_KEY; A credential in JUDGE_API_KEY.

Does LLM Benchmark access the network?

SKILL.md contains no URLs. Its commands use uv, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is LLM Benchmark safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does LLM Benchmark use?

LLM Benchmark is published under the GPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does LLM Benchmark use?

About 1.6k tokens (SKILL.md is roughly 6.4k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to LLM Benchmark?

Skills that share tags, products or a category with LLM Benchmark: Asd Ste100 (danyuchn/asd-ste100-skill, 4.3k stars), Summarize Anything (swyxio/skills, 176 stars), Technical Writing Standard (cursor/plugins, 11k stars) and Ponytail Audit (DietrichGebert/ponytail, 160k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains LLM Benchmark?

lumose-health (a GitHub organization) maintains it in lumose-health/GlycemicGPT, which has 141 GitHub stars. The repository holds 3 skills in this directory. The repository was last updated on October 8, 2026.

Source: lumose-health/GlycemicGPT on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.

Install LLM Benchmark

Facts

At a glance

What it does

When your agent uses it

Example prompts

Requirements

Workflow steps

What it can do on your machine

Tool permissions

Runs code

Network

Credentials

Context cost

Safety

SKILL.md

llm-benchmark skill

Step 1 — Confirm the model under test

Step 2 — Run the suites

Step 3 — Read each JSON report and apply the hard rule

Step 4 — Translate into a plain verdict

Step 5 (optional) — Extensions

Constraints

Files

Compare with similar skills

Similar skills

More from lumose-health/GlycemicGPT

Related

Questions about LLM Benchmark