is-cli-agentic

cold CLI trial · no setup · no credentials

Can an agent finish the task?

Enter a CLI. We publish the checks, three identical agent trials, and the raw trajectory—whether it passes or not.

run a public scan$ is-cli-agentic
100 launch CLIs3+ trials before rank9 dimensions, N/A is free1 public trajectory per verdict

public evidence ledger

100 CLIs, one score each, ranked high to low.

22ranked
observed facts
rankCLI / packagescore
1bun
pkgbrew/oven-sh/bun/bun
70.0%
2railway
cloudbrew/railway
69.2%
3rg
utilbrew/ripgrep
65.2%
4pnpm
pkgnpm/pnpm
64.2%
5pip3
pkgbuiltin/pip
61.7%
6psql
databrew/postgresql@18
60.8%
7curl
utilbuiltin/curl
60.0%
8jq
utilbrew/jq
59.6%
9auth0
saasbrew/auth0/auth0-cli/auth0
57.8%
10git
vcsbuiltin/git
56.7%
11vercel
cloudnpm/vercel
55.4%
12claude
agentnpm/@anthropic-ai/claude-code
54.2%
13codex
agentnpm/@openai/codex
53.3%
14gh
vcsbinary/cli/cli
52.5%
15npm
pkgbuiltin/npm
51.7%
16docker
containerbrew/docker
50.8%
17kubectl
containerbinary/kubernetes/kubernetes
50.0%
18supabase
cloudbrew/supabase/tap/supabase
48.3%
19sqlite3
databrew/sqlite
46.2%
20neon
datanpm/neonctl
38.3%
21wrangler
cloudnpm/wrangler
38.3%
22opencode
agentnpm/opencode-ai
30.0%

coming soon · capture queued

Score is the agent-verified pass rate once three trials exist, or the deterministic facts score before that. Every CLI links through to the evidence behind its number.