Cultivar

CLI that measures whether an agent skill actually beats a no-skill baseline

Works with setup

Test report

Verdict
Works with setup
Score
8.0/10
Tested
Jul 18, 2026
Environment
Claude Code 2.x (agent harness)
Upstream re-checked
Jul 30, 2026 · 8368e5b

This skill is no longer available upstream. Our re-check on Aug 10, 2026 couldn't find it any more (repo unreachable/deleted). The test below is what we measured on Jul 18, 2026 and we're leaving it up as a record — but there is nothing left to install, so we've removed the command.

Installed the CLI with uv and its packaged 'cultivar hello' smoke task ran end-to-end against the real Claude runner, writing per-run JSON, Markdown, and JSONL traces plus a workdir exactly as documented. The harness itself is solid, but the actual payoff (a graded with-skill vs without-skill comparison) needs an authed agent CLI plus an ANTHROPIC_API_KEY for the grader; here the runner returned a 401 and grading was skipped, so the headline delta stays unmeasured. The skill is a faithful, accurate command reference over a genuinely working tool. Real setup friction, a paid key plus a logged-in runner, sits between install and any result.

Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works

  • Installs cleanly 5/5
  • Triggers reliably 5/5
  • Output vs. baseline 5/10
  • Docs & honesty 5/5

What Cultivar does

Drives the cultivar CLI to evaluate whether an agent skill improves behavior: scaffolds tasks, runs each with-skill, without-skill, and with-docs across Claude/Copilot/Gemini locally or in Modal sandboxes, then an LLM grader scores runs against a rubric. Triggers when a user wants to create, run, or interpret cultivar skill evals.

How to install Cultivar

Nothing to install: the source repository no longer has this skill. If the author brings it back, our daily re-check will pick it up and the command will reappear here.

Commands — how to trigger Cultivar

  • /cultivar CLI that measures whether an agent skill actually beats a no-skill baseline

It also activates on plain-language prompts like these:

  • Does this skill I wrote actually help, can you eval it properly?
  • Run this skill with and without it enabled, then grade the results.
  • Help me interpret these eval results comparing skill performance.

Frequently asked questions

Is the Cultivar skill free?
Yes. The skill itself is free from pinecone-io/cultivar. SkillProof publishes the install command and an independent test verdict at no cost.
Does Cultivar work with Claude Code?
We tested it with Claude Code 2.x (agent harness) on Jul 18, 2026. Verdict: Works with setup. Installed the CLI with uv and its packaged 'cultivar hello' smoke task ran end-to-end against the real Claude runner, writing per-run JSON, Markdown, and JSONL traces plus a workdir exactly as documented. The harness itself is solid, but the actual payoff (a graded with-skill vs without-skill comparison) needs an authed agent CLI plus an ANTHROPIC_API_KEY for the grader; here the runner returned a 401 and grading was skipped, so the headline delta stays unmeasured. The skill is a faithful, accurate command reference over a genuinely working tool. Real setup friction, a paid key plus a logged-in runner, sits between install and any result.
What is the Cultivar SkillProof Score?
8.0/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 5/5.
How do I install Cultivar?
Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
Can I use Cultivar with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.