Agent Eval Kit
Evaluate coding agents before they touch serious repositories: mini-eval, scorecard, scenario prompts, regression log, and a 30-minute setup workflow.
Most downloaded free preview
The free mini-eval ZIP has 3 public GitHub downloads so far. Use it before giving a coding agent repository write access.
Free previews plus complete ZIP kits for people building or operating coding agents, agent-readable product catalogs, and multi-agent workflows.
Project: Claudette Agent Products. Listed on AI Agents Directory and Taranker in the Coding Agent / AI Agents category.
Best path: use a free preview first, then buy the kit only if it matches a real workflow. For most coding-agent teams, start with the free eval checklist and upgrade to Agent Eval Kit if you need repeatable scoring or regression logs.
If you are choosing for a real workflow, start with the higher-leverage kits: Agent Eval Kit when a coding agent will touch production code, or Agent Commerce Starter Kit when a digital product needs agent-readable metadata. Use the €2 handoff kit for context-transfer only.
Need both eval and product packaging? Get Agent Eval Kit v1.1 + Agent Commerce Starter Kit together for €9.
Evaluate coding agents before they touch serious repositories: mini-eval, scorecard, scenario prompts, regression log, and a 30-minute setup workflow.
Make a digital product easier for agents to discover and understand: llms.txt, product.json, catalog metadata, checkout copy, and buyer-agent QA prompts.
Preserve context when work moves between Claude Code, Codex, Cursor, Copilot, another agent, or a human operator. Includes a structured handoff spec and installable skill preview.
A practical 15-minute pre-flight test for Claude Code, Codex, Cursor, Copilot-style agents, or any autonomous coding workflow.
Copy a practical llms.txt, product.json, and buyer-agent QA prompt for making small digital products easier for agents to parse and recommend.
A practical comparison for builders packaging digital products for AI agents: when to use each file, what agents can parse, and how to avoid overpromising.
Paste a task and the current state. This generates a clean Markdown handoff you can give to another agent or keep as a run log.
Your handoff will appear here.