Can an AI agent actually buy from your store? Score it, see the evidence, fix the gaps.
19 MCP servers
The Instagram adapter, formalized: capability-declared, hash-bound native variants for Feed
The insurance watcher — signals your policies' expiries before they happen. It does not insure
One approved Task Brief, every marketplace: hash-bound briefs, all-in quotes, and a hire that runs
The governed, vendor-neutral VM fleet layer for agents: one policy, one receipt ledger, every cloud
Drafts once, syndicates drafts everywhere sanctioned, unifies the analytics. Never publishes
Signed Agent Cards, policy-gated delegation, certification before hire.
The Customer Graph core: one register of the people an account serves, with provenance on every
One canonical, hash-bound package behind every Eclipse social adapter. It publishes nothing itself
Discover agent tools by problem: every listing carries its Test Bench verdict or says it has none.
Agent birth-certificate sealing + out-of-purpose detection, fixture records; detects, never blocks.
Point an agent at a codebase: narrowed-scope architecture map, dependency/EOL risk list.
Governed design-to-print authority for agents — signed provenance from model to machine, safety
Is this bench result real? Who issued it, at what scope, anchored where? One call answers.
Warranty witness for agent purchases: sealed, transferable claim packs. LOCAL/fixture only.
ACME certificate lifecycle for agents: order, HTTP-01/DNS-01 validation, download, renewal.
Policy-scoped agent wallet: caps, recipient allowlist, human approvals, kill switch, audit trail.
Signed agent-only friction board: findings backed by verified trace packs, in a closed vocabulary.
MCP deprecation radar: witnessed, hash-chained registry snapshots prove deprecated/removed tools.