Borrowing it
Nothing to install: this file belongs to timothywarner-org/claude-code. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/timothywarner-org/claude-code/main/.claude/skills/genai-prompt-eval/SKILL.mdgit clone --depth 1 https://github.com/timothywarner-org/claude-codeWrote 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/timothywarner-org/claude-code/genai-prompt-eval)<a href="https://agentmods.dev/skills/timothywarner-org/claude-code/genai-prompt-eval"><img src="https://agentmods.dev/badge/skills/timothywarner-org/claude-code/genai-prompt-eval.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 5.1 | $0.00103 | $0.00631 |
| Opus 5 | $0.00051 | $0.00316 |
| Sonnet 5 | $0.00021 | $0.00126 |
| Haiku 4.5 | $0.00010 | $0.00063 |
Grade A, and why
genai-prompt-eval 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 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.
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 — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Score GenAI outputs before shipping
This skill measures whether a generative-AI feature produces answers that are grounded, relevant, coherent, and safe. It runs a set of eval cases through the model, scores each output on those four dimensions, and reports pass or fail against thresholds. Pair it with the azure-ai-deploy skill: evals are Gate 1 of that deploy checklist.
When to reach for this
- A GenAI feature is changing and you need a regression signal on answer quality.
- A deploy gate requires proof that outputs meet a quality bar.
- You want a repeatable eval corpus that reflects real enterprise questions, not toy prompts.
Workflow
1. Read the dimensions
Read resources/references/EVAL-DIMENSIONS.md. It defines groundedness, relevance, coherence, and safety, states what each one measures, and gives a pass signal for each.
2. Build the eval corpus
Start from resources/templates/eval_cases.jsonl. Each line is one case: an input prompt, optional context the answer must stay grounded to, and expected_criteria describing a passing answer. Add cases that mirror the questions real users send.
3. Run the harness
uv run python ${CLAUDE_SKILL_DIR}/resources/scripts/run_eval.py \
--cases ${CLAUDE_SKILL_DIR}/resources/templates/eval_cases.jsonl \
--threshold 0.8
The script loads the cases, calls the model for each, scores the output on the four dimensions, prints a per-case and aggregate report, and exits non-zero when the aggregate score falls below the threshold. That non-zero exit fails a CI or deploy step.
4. Read the report and act
- Cases below threshold name the failing dimension. Fix the prompt, the retrieval context, or the guardrail, then re-run.
- Record the aggregate score as the new baseline so the next run detects regressions.
Conventions
- uv manages Python, not pip. Run scripts with
uv run. - No hardcoded secrets. The scoring model client reads its endpoint and deployment from env vars.
- Realistic cases only. Eval inputs are enterprise scenarios, never placeholder prompts.
- Deterministic scoring where possible. Prefer a low temperature on any model-graded dimension so scores are stable across runs.
What ships with it
3 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 · 48 lines · 103 tokens per session scan A d12cf865df8a
genai-prompt-eval is a skill published in the GitHub repository timothywarner-org/claude-code (223 stars, last pushed 1mo ago), licensed MIT. It adds 103 tokens to every session and 631 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
cli-eval
Create and run evaluation suites, watch live benchmark progress, view scorecards, compare model performance, and integrate eval runs with CI workflows from the CLI.
model-merging
Merge multiple fine-tuned models using mergekit to combine capabilities without retraining. Use when creating specialized models by blending domain-specific expertise (math + coding + chat), improving performance beyond single models, or experimenting rapidly with model variants. Covers SLERP, TIES-Merging, DARE, Task…
darwinian-evolver
Evolve prompts/regex/SQL/code with Imbue's evolution loop.
validate
Validate Semantica pipelines, extraction quality, graph schemas, and ontology consistency. Returns structured error/warning checklists. Uses PipelineValidator, PipelineBuilder.validatepipeline(), GraphValidator, and OntologyValidator. Sub-commands: pipeline, step, dependencies, extraction, graph, ontology, performance.
launching-evals
Run, monitor, analyze, and debug LLM evaluations via nemo-evaluator-launcher. Covers running evaluations, checking status and live progress, debugging failed runs, exporting artifacts and logs, and analyzing results. ALWAYS triggers on mentions of running evaluations, checking progress, debugging failed evals…
nemo-automodel-recipe-development
Create and modify NeMo AutoModel training and evaluation recipes, including YAML structure, builders, and execution flow.