challenge-run

Execute predefined challenges against an agent to evaluate performance and confidence calibration.

14|6|Updated Feb 3, 2026
One-click install
npx skills add https://github.com/tyevans/tackline --skill challenge-run
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: challenge-run
Source: https://github.com/tyevans/tackline/tree/main/skills/workflows/challenge-run
Command: npx skills add https://github.com/tyevans/tackline --skill challenge-run

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill automates the execution and rigorous evaluation of agent challenges, providing objective performance metrics and insights into agent capabilities.

Core Features & Use Cases

  • Automated Challenge Execution: Dispatches agents against predefined challenge sets to test their performance.
  • Performance Evaluation: Assesses agent output against acceptance criteria, hidden traps, and ground truth.
  • Calibration & Calibration: Measures how well an agent's self-reported confidence aligns with its actual performance.
  • Use Case: After generating a set of challenges for your billing-agent using /challenge-gen, you can run /challenge-run billing-agent to see how effectively it handles those specific scenarios and identify areas for improvement.

Quick Start

Use the challenge-run skill to execute challenges for the agent named 'billing-agent'.

Frequently Asked Questions about challenge-run

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I evaluate agent performance against predefined challenges?

Agent performance evaluation is automated by dispatching agents against predefined challenge sets to test output, trap detection, and confidence calibration. It measures how well self-reported confidence aligns with actual performance metrics.

What is confidence calibration in agent evaluation?

Confidence calibration in agent evaluation measures the alignment between an agent's self-reported confidence and its actual performance results. It provides objective metrics to determine if an agent accurately assesses its own reliability.

How do I run edge-case and commit-replay challenge strategies?

Edge-case and commit-replay challenge strategies are executed by dispatching agents against predefined challenge files. The commit-replay strategy specifically leverages worktree isolation to safely replay scenarios during agent evaluation.

Do I need challenge files and team.yaml definitions to execute agent challenges?

Yes, executing agent challenges requires predefined challenge files and agent definitions from team.yaml. These inputs are necessary for comprehensive assessment of performance, trap detection, and calibration metrics.

What is the best way to automate agent challenge execution and testing?

Automating agent challenge execution involves dispatching specified agents against predefined challenge sets to assess performance, hidden traps, and ground truth acceptance criteria. It provides objective metrics for identifying areas of improvement.

Why are my agent challenge evaluations not detecting hidden traps?

Agent challenge evaluations detect hidden traps by assessing agent output against predefined acceptance criteria and ground truth. Ensure your challenge files and team.yaml agent definitions are properly configured for comprehensive assessment.