agentwares
Small products that sell to AI-native startups and to their agents. Every product has an MCP
server, an llms.txt, a machine-readable pricing.json, and a first
call that needs no signup.
agentguard
MCP policy proxy and SDK middleware: hard spend limits, destructive-action approvals, a kill switch, scoped keys, dry-run diffs and a tamper-evident audit log. Open source, no account. See a live run.
npx @agentwares/agentguardmcpcheck
Nightly checks for people who publish MCP servers: clean install on Windows, macOS and Linux, a client-version matrix, OAuth conformance and tool-catalog drift, with a badge. Sample report.
$19 / $49 per serveragentcheck
We re-run your suite when a model changes under you, and tell you what broke. Uptime checks and a public badge are free, and so is the daily model-drift index.
free / $29 / $199 / $499morning digest
One 7am email for every agent that ran overnight, across Claude Code, Codex, Devin and Cursor: what merged, what it cost, what failed, what needs a decision. Sample.
$15/mo · the runner is freeshelf
Per-call tools an agent can buy without a subscription: a real-estate deal memo from an address, and stablecoin reserve attestations as JSON. Sample calls are free.
$0.05–$0.75 per callfree tools
Can an agent sign up to your SaaS? We try, and publish the evidence. Plus a linter that checks whether your AGENTS.md and CLAUDE.md still match your repo.
free