LLM Benchmarking with lm-evaluation-harness
Orchestra-Research/AI-Research-SKILLs
Runs lm-evaluation-harness to benchmark language models on academic suites such as MMLU, GSM8K and HumanEval, compare models and track training checkpoints.
Run evaluation tests for prompt quality. An agent skill from Azure/azure-sdk-tools.
Install Run Evals
$ npx skills add Azure/azure-sdk-tools --skill run-evals -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install Azure/azure-sdk-tools run-evals --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/Azure/azure-sdk-tools.git skills-src && mkdir -p .claude/skills && cp -r skills-src/packages/python-packages/apiview-copilot/.github/skills/run-evals .claude/skills/run-evals && rm -rf skills-srcUse ~/.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/
Install the "run-evals" agent skill from https://github.com/Azure/azure-sdk-tools/tree/main/packages/python-packages/apiview-copilot/.github/skills/run-evals into .claude/skills/run-evals/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "run-evals", 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 install https://github.com/Azure/azure-sdk-tools/tree/main/packages/python-packages/apiview-copilot/.github/skills/run-evalsType 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.
$ npx skills add Azure/azure-sdk-tools --skill run-evals -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install Azure/azure-sdk-tools run-evals --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/Azure/azure-sdk-tools.git skills-src && mkdir -p .agents/skills && cp -r skills-src/packages/python-packages/apiview-copilot/.github/skills/run-evals .agents/skills/run-evals && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "run-evals" agent skill from https://github.com/Azure/azure-sdk-tools/tree/main/packages/python-packages/apiview-copilot/.github/skills/run-evals into .agents/skills/run-evals/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "run-evals", 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.
$ npx skills add Azure/azure-sdk-tools --skill run-evals -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install Azure/azure-sdk-tools run-evals --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/Azure/azure-sdk-tools.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/packages/python-packages/apiview-copilot/.github/skills/run-evals .cursor/skills/run-evals && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "run-evals" agent skill from https://github.com/Azure/azure-sdk-tools/tree/main/packages/python-packages/apiview-copilot/.github/skills/run-evals into .cursor/skills/run-evals/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "run-evals", 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 https://github.com/Azure/azure-sdk-tools.git --path packages/python-packages/apiview-copilot/.github/skills/run-evals--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add Azure/azure-sdk-tools --skill run-evals -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install Azure/azure-sdk-tools run-evals --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/Azure/azure-sdk-tools.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/packages/python-packages/apiview-copilot/.github/skills/run-evals .gemini/skills/run-evals && rm -rf skills-srcUse ~/.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/
Install the "run-evals" agent skill from https://github.com/Azure/azure-sdk-tools/tree/main/packages/python-packages/apiview-copilot/.github/skills/run-evals into .gemini/skills/run-evals/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "run-evals", 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.
$ gh skill install Azure/azure-sdk-tools run-evalsInstalls 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).
$ npx skills add Azure/azure-sdk-tools --skill run-evals -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/Azure/azure-sdk-tools.git skills-src && mkdir -p .github/skills && cp -r skills-src/packages/python-packages/apiview-copilot/.github/skills/run-evals .github/skills/run-evals && rm -rf skills-srcUse ~/.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/
Install the "run-evals" agent skill from https://github.com/Azure/azure-sdk-tools/tree/main/packages/python-packages/apiview-copilot/.github/skills/run-evals into .github/skills/run-evals/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "run-evals", 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.
$ npx skills add Azure/azure-sdk-tools --skill run-evals -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install Azure/azure-sdk-tools run-evals --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/Azure/azure-sdk-tools.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/packages/python-packages/apiview-copilot/.github/skills/run-evals .opencode/skills/run-evals && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "run-evals" agent skill from https://github.com/Azure/azure-sdk-tools/tree/main/packages/python-packages/apiview-copilot/.github/skills/run-evals into .opencode/skills/run-evals/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "run-evals", 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
run-evals- GitHub stars
- 134
- Token cost
- ~977 tokens
- SKILL.md length
- 271 words
- Files
- 1
- Skills in repo
- 35
- Repo updated
- First seen
- Licence
- MIT
At a glance
Run evaluation tests for prompt quality. An agent skill from Azure/azure-sdk-tools.
- Validate prompt changes
- SKILL.md covers When to Use, Running Evals, Existing Workflows and Recordings, plus 1 more section
- Calls python
- Check prompt quality
What it does
Run Evals is an agent skill from Azure/azure-sdk-tools, published by the product's own GitHub organization. Run evaluation tests for prompt quality. Use for: run evals, validate prompt changes, check prompt quality, eval recording, evaluation failure, debug eval, use recording.
Its SKILL.md is about 980 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 AI & LLM Engineering, covering LLM evaluation. The repository describes itself as: Tools repository leveraged by the Azure SDK team. The licence is MIT.
When your agent uses it
- Validate prompt changes
- Check prompt quality
- Evaluation failure
Example prompts
- “/run-evals”
Requirements
- Python 3
What it can do on your machine
Read from SKILL.md and the folder at commit 6e4fb2b. 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:
python
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Run Evals loads about 977 tokens when it runs. Until then it costs about 45 tokens; SKILL.md has 271 words of instructions outside code blocks.
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 Azure/azure-sdk-tools at commit 6e4fb2b, republished under its MIT licence (© Azure). 271 words, ~977 tokens.
Download SKILL.mdSave it as
.claude/skills/run-evals/SKILL.md (or your agent's skills folder).- name
- run-evals
- description
- Run evaluation tests for prompt quality. Use for: run evals, validate prompt changes, check prompt quality, eval recording, evaluation failure, debug eval, use recording.
Run Evaluations
When to Use
- After modifying any
.promptyfile to validate quality hasn't regressed - Debugging failing or partial eval results
- Iterating on prompts with cached recordings to avoid LLM costs
Running Evals
Always activate the virtualenv first: .venv\Scripts\activate (Windows) or source .venv/bin/activate (Linux/macOS).
Via CLI
# Run all workflows
avc test eval
# Run a specific workflow
avc test eval --test-paths evals/tests/mention_action
# Run a single test file
avc test eval --test-paths evals/tests/filter_existing_comment/discard_azure_sdk_repeat_comment.yaml
# Multiple runs (median result kept)
avc test eval --num-runs 5 --test-paths evals/tests/filter_comment_metadata
# Use recordings (cached LLM responses) — first run saves, subsequent runs reuse
avc test eval --use-recording --test-paths evals/tests/mention_action
# Verbose output (show passing tests too)
avc test eval --style verboseVia run.py directly
cd evals
python run.py --test-paths tests/mention_actionExisting Workflows
| Workflow directory | Kind | Target function in _custom.py | Prompt tested |
|---|---|---|---|
mention_action | prompt | _mention_action_workflow | parse_conversation_action.prompty |
mention_summarize | summarize_prompt | _mention_summarize_workflow | summarize_github_actions.prompty |
thread_resolution_action | prompt | _thread_resolution_action_workflow | parse_thread_resolution_action.prompty |
filter_comment_metadata | prompt | _filter_comment_metadata | filter_comment_with_metadata.prompty |
filter_existing_comment | prompt | _filter_existing_comment | filter_existing_comment.prompty |
deduplicate_parser_issue | prompt | _deduplicate_parser_issue | deduplicate_parser_issue.prompty |
deduplicate_guidelines_issue | prompt | _deduplicate_guidelines_issue | deduplicate_guidelines_issue.prompty |
filter_generic_comment | prompt | _filter_generic_comment | filter_generic_comment.prompty |
judge_comment_confidence | prompt | _judge_comment_confidence | judge_comment_confidence.prompty |
merge_comments | summarize_prompt | _merge_comments | merge_comments.prompty |
generate_correlation_ids | prompt | _generate_correlation_ids | generate_correlation_ids.prompty |
Evaluator kinds
prompt— Action-based. Compares expected vs actual action, then similarity-scores the rationale. Wrong action = 0%.summarize_prompt— Summary-based. UsesSimilarityEvaluatoron full output. Success threshold: score > 70%.
Recordings
- Stored in
evals/recordings/<workflow_name>/<testcase_id>.json - Gitignored — each dev builds their own cache
- If you change a test file, delete its recording or run without
--use-recording --use-recordingon first run makes LLM calls and saves; subsequent runs reuse cached responses
Gotchas
- Use
python cli.pynot.\avc: Theavc.batscript calls barepythonwhich may resolve to the system Python instead of the venv. Use.venv\Scripts\activate; python cli.py test eval ...to ensure the venv Python is used. - Field name mismatch: Test YAML fields must exactly match target function parameter names (excluding
testcaseandresponse) - Stale recordings: After changing a prompt, delete recordings or run without
--use-recordingto get fresh results - Testcase uniqueness: The
testcasefield must be unique across all test files in a workflow — it's the cache key - Kind validation: The
kindintest-config.yamlmust be registered in_config_loader.py(promptorsummarize_prompt)
© Azure, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
Just SKILL.md in packages/python-packages/apiview-copilot/.github/skills/run-evals of Azure/azure-sdk-tools.
Open the folder on GitHubat commit 6e4fb2b
Compare with similar skills
Run Evals 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.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Run Evals this skillAzure/azure-sdk-tools | 134 | — | ~977 | Automated safety check: Pass | MIT | |
| LLM Benchmarking with lm-evaluation-harnessOrchestra-Research/AI-Research-SKILLs | 13k | 8 repos | ~3k | Automated safety check: Pass | MIT | |
| Azure AI Projects Python SDKmicrosoft/skills | 3.1k | 6 repos | ~2.8k | Automated safety check: Pass | MIT | |
| Fine-Tuning ExpertJeffallan/claude-skills | 12k | 1 repos | ~1.7k | Automated safety check: Pass | MIT | |
| Looperksimback/looper | 710 | — | ~2.7k | Automated safety check: Notes | MIT | |
| Hugging Face Local Model Evalshuggingface/skills | 11k | 2 repos | ~1.6k | Automated safety check: Pass | Apache-2.0 |
Similar skills
- Official
Azure AI Projects Python SDK
microsoft/skills
Reference for building on Microsoft Foundry with the azure-ai-projects Python SDK: project clients, versioned agents, evaluations, connections, datasets and indexes.
Fine-Tuning Expert
Jeffallan/claude-skills
Guides LLM fine-tuning with LoRA and QLoRA through Hugging Face PEFT, from dataset validation and training checks to adapter merging, quantization and deployment.
Looper
ksimback/looper
Scaffold a well-designed agent loop with best-practice coaching and a cross-model review council.
- Official
Hugging Face Local Model Evals
huggingface/skills
Runs evaluations of Hugging Face Hub models on local hardware with inspect-ai or lighteval, and helps choose between vLLM, Transformers and accelerate backends.
- Official
Agent Eval Engineering
langchain-ai/langchain-skills
Builds agent evaluations in stages: inspect the repository and traces, agree a Task Spec with you, then build, audit and run a Harbor task with an independent verifier.
More from Azure/azure-sdk-tools
All 35 skills in this repo- Official
Apiview Feedback Resolution
Azure/azure-sdk-tools
Analyze and resolve APIView review feedback on Azure SDK PRs.
- Official
Azsdk Common Apiview Feedback Resolution
Azure/azure-sdk-tools
Analyze and resolve APIView review feedback on Azure SDK PRs.
- Official
Azsdk Common Live And Recorded Tests
Azure/azure-sdk-tools
Deploy test resources and run Azure SDK tests in live, record, or playback mode.
- Official
Azsdk Common Pipeline Analysis
Azure/azure-sdk-tools
Analyze Azure SDK CI/CD pipeline failures into a structured diagnosis, and define the required output format.
- Official
Azsdk Common Prepare Release Plan
Azure/azure-sdk-tools
Create, get, update, abandon, and link SDK PRs to release plan work items for Azure SDK releases.
- Official
Azure Typespec Assessment
Azure/azure-sdk-tools
Assess Azure TypeSpec Git diffs for semantic intent, REST and downstream SDK breaking changes, Azure Guidelines compliance, and documentation completeness.
Related
Categories
Questions about Run Evals
What does Run Evals do?
Run evaluation tests for prompt quality. An agent skill from Azure/azure-sdk-tools. Run Evals is an agent skill from Azure/azure-sdk-tools, published by the product's own GitHub organization. Run evaluation tests for prompt quality.
When should I use Run Evals?
Run Evals fits situations like: validate prompt changes; check prompt quality; evaluation failure.
How do I install Run Evals in Claude Code?
Run `npx skills add Azure/azure-sdk-tools --skill run-evals -a claude-code`. Or copy the skill folder (packages/python-packages/apiview-copilot/.github/skills/run-evals in Azure/azure-sdk-tools) into .claude/skills/run-evals in your project. Claude Code loads it when a task matches its description.
How do I install Run Evals in Codex?
Run `npx skills add Azure/azure-sdk-tools --skill run-evals -a codex`. Or copy the skill folder (packages/python-packages/apiview-copilot/.github/skills/run-evals in Azure/azure-sdk-tools) into .agents/skills/run-evals in your project. Codex loads it when a task matches its description.
Can I use Run Evals 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 Azure/azure-sdk-tools --skill run-evals -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/run-evals, .gemini/skills/run-evals, .github/skills/run-evals and .opencode/skills/run-evals in your project.
What does Run Evals need to run?
Going by SKILL.md and its folder, Run Evals needs the command-line tools its instructions call (python). Our summary lists: Python 3.
Does Run Evals access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Run Evals 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 Run Evals use?
Run Evals is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Run Evals use?
About 977 tokens (SKILL.md is roughly 3.9k 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 Run Evals?
Skills that share tags, products or a category with Run Evals: LLM Benchmarking with lm-evaluation-harness (Orchestra-Research/AI-Research-SKILLs, 13k stars), Azure AI Projects Python SDK (microsoft/skills, 3.1k stars), Fine-Tuning Expert (Jeffallan/claude-skills, 12k stars) and Looper (ksimback/looper, 710 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Run Evals?
Azure (a GitHub organization, an official publisher) maintains it in Azure/azure-sdk-tools, which has 134 GitHub stars. The repository holds 35 skills in this directory. The repository was last updated on October 8, 2026.
Source: Azure/azure-sdk-tools on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.