agent-evaluation¶

Target Audience: Founders, operators, and team leads evaluating AI agent tools for knowledge work delegation.

Area: Foundational
Category: utilities
Level: beginner
Input: conversation
Output: document
Setup: none
Roles: CEO / Founder, Product Manager, Operations Manager

Evaluate AI agent tools against three structural dimensions (Persistent Memory, Inspectable Surfaces, Compounding Context) and generate calibrated delegation specs.

How you ask¶

“Evaluate this agent”
“Should I use this tool”
“Agent comparison”

Inputs¶

tool-name — Name of the agent tool being evaluated
task-description — Specific task or workflow intended for delegation
quality-definition — What 'good' looks like for this task
tools — Comma-separated list of tool names to compare
task-description — The task all tools will be evaluated against

Outputs¶

comparison (markdown)
scorecard (markdown)
{company.decisions}/agent-comparison-{YYYY-MM-DD}.md
{company.decisions}/agent-eval-{tool-slug}-{YYYY-MM-DD}.md