pauloribeiro16pauloribeiro16/my-skills/eval-runner/SKILL.md

eval-runner

Skill1 starsChanged 6 months agoMITView on GitHub

Use when running evaluation tasks. Ensures progressive eval (1 task → 2-3 → max 5), proper trial count, and correct result interpretation. Prevents timeout issues with large batches.

1 other project uses a version of this.

Adopted by

Most recently taken up by

MarchRory/liushi_agent_settingstheir own version, first seen Jun 1, 2026

Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
MarchRory/liushi_agent_settings★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository41 other files it gives its agents.
- AGENTS.mdAGENTS.md
- academic-evaluationSkill
- academic-writingSkill
- agent-orchestrationSkill
- agents-md-writerSkill
- algorithmic-artSkill
- brand-guidelinesSkill
- codebase-architectureSkill
- code-reviewSkill
- commit-workflowSkill
- context-checkpointSkill
- doc-coauthoringSkill
- docxSkill
- etl-runnerSkill
- feature-devSkill
- frontend-designSkill
- hpc-deucalionSkill
- internal-commsSkill
- langchainSkill
- langfuseSkill
- langgraphSkill
- llm-observability-stackSkill
- mcp-builderSkill
- neo4j-verifySkill
- pdfSkill
- pptxSkill
- project-conventionsSkill
- python-best-practicesSkill
- react-namingSkill
- security-hooksSkill
- service-verifySkill
- skill-creatorSkill
- skill-inspectorSkill
- slack-gif-creatorSkill
- sprint-contractSkill
- theme-factorySkill
- uml-modelingSkill
- webapp-testingSkill
- web-artifacts-builderSkill
- web-debugSkill
- xlsxSkill

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 7 are distinctive, and 1 is very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Distinctive:
When checking eval results
line 14
Distinctive:
When troubleshooting low scores
line 15
Very specific:
Remove --task flag to run all tasks in task_bank.yaml
line 46
Distinctive:
Pre-check: Verify ETL ran successfully
line 81
Distinctive:
If pass: Run 2-3 more tasks
line 83
Distinctive:
If still passing: Run up to 5
line 84
Distinctive:
Always start with 1 task
line 129
Distinctive:
If 3 consecutive timeouts: STOP and report
line 130

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

SKILL.md

Eval Runner Skill

Runs evaluation tasks correctly using progressive batching. Prevents Ollama/Minimax timeout issues.

When to Activate

Running any evaluation task
Before executing run_eval.py
When checking eval results
When troubleshooting low scores

Progressive Eval Rule (CRITICAL)

NEVER run all tasks at once.

Phase	Tasks	Why
Step 1	1 task	Verify setup works
Step 2	2-3 tasks	Check consistency
Step 3	Max 5 tasks	Standard batch
Full suite	Only with explicit approval	Timeout risk

Command Templates

Single Task (always start here)

PYTHONPATH=. python core/eval/run_eval.py \
  --case [case_path] \
  --task T001 \
  --trials 1 \
  --verbose

Small Batch (2-5 tasks)

PYTHONPATH=. python core/eval/run_eval.py \
  --case [case_path] \
  --trials 1 \
  --verbose

Remove --task flag to run all tasks in task_bank.yaml

Single Task with Specific ID

PYTHONPATH=. python core/eval/run_eval.py \
  --case [case_path] \
  --task T017 \
  --trials 1 \
  --verbose

Environment Setup

# Activate venv
source /path/to/venv/bin/activate

# Set PYTHONPATH
export PYTHONPATH=.

# Verify case config
cat [case_path]/case.yaml | grep -E "neo4j|ollama"

Understanding Results

Score	Meaning	Action
0.0	No data loaded OR wrong ports	Check ETL, check ports
0.5-0.7	Partial match	Review expected vs actual
0.8-1.0	Good match	Task passed
timeout	Model not responding	Reduce batch size

Evaluation Flow

Pre-check: Verify ETL ran successfully
Run single task: T001 --trials 1
If pass: Run 2-3 more tasks
If still passing: Run up to 5
If fail: Debug before continuing

Trial Count

Situation	Trials
Quick smoke test	1
Standard evaluation	1
Investigating intermittent failure	3
Final validation	1

Troubleshooting Low Scores

Symptom	Check
All scores 0.0	ETL ran? Data in Neo4j?
Some scores 0.0	Specific tasks failing?
Timeout	Batch too large? Model overloaded?
Wrong answers	case_path correct? case property set?

Output Format

## Eval Results

**Case:** [case_name]
**Tasks run:** [N]
**Passed:** [M]
**Failed:** [K]

### Per-Task Results
| Task | Score | Notes |
|------|-------|-------|
| T001 | 1.0 | PASS |
| T002 | 0.0 | FAIL — no data |
| T003 | 0.8 | Partial |

**Next step:** [recommendation]

Safety Limits

Max 5 tasks per batch
Max 3 trials per task
Always start with 1 task
If 3 consecutive timeouts: STOP and report

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free