Eval

This is a basic code evaluation skill that executes Python code in a sandboxed environment and returns the output. You'd reach for this when Claude needs to run calculations, test logic, or process data that's easier to handle programmatically than through conversation. The implementation is straightforward: it takes code as input, runs it, and gives you back whatever gets printed or returned. One thing to note is that this is really foundational stuff. It's less a specialized tool and more a primitive capability that other, more complex skills probably build on top of. If you're debugging why Claude can or can't execute something, this is where that ability comes from.

skills.sh installs: 1.4K
GitHub stars: 29.5K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add alirezarezvani/claude-skills --skill eval --agent claude-code

Installs into .claude/skills of the current project.

Files

/agenthub:eval — Evaluate Agent Results

Rank all agent results for a session. Supports metric-based evaluation (run a command), LLM judge (compare diffs), or hybrid.

Usage

/agenthub:eval                           # Eval latest session using configured criteria
/agenthub:eval 20260317-143022           # Eval specific session
/agenthub:eval --judge                   # Force LLM judge mode (ignore metric config)

What It Does

Metric Mode (eval command configured)

Run the evaluation command in each agent's worktree:

python {skill_path}/scripts/result_ranker.py \
  --session {session-id} \
  --eval-cmd "{eval_cmd}" \
  --metric {metric} --direction {direction}

Output:

RANK  AGENT       METRIC      DELTA      FILES
1     agent-2     142ms       -38ms      2
2     agent-1     165ms       -15ms      3
3     agent-3     190ms       +10ms      1

Winner: agent-2 (142ms)