Covenant
CVNT #6540Project Safety Score: 26%
Updated 1 day, 5 hours ago& Market
Credibility
Behavioral
Mechanics
Distribution
Covenant is open infrastructure for AI agents: an operating layer that sits below agent applications and above the host OS, providing the controls autonomous agents need to operate safely. It exposes scoped, cryptographically signed capabilities that gate what an agent may do; a budget ledger that caps how much it may spend; durable tiered memory; runtime isolation; MCP, HTTP, and agent-to-agent interfaces; and an append-only, hash-chained audit log across all operations.
$CVNT is the settlement credit of the network. Every resource an agent consumes — memory writes, tool calls, external API requests, messages — is metered as a signed receipt priced in credits. Credits are minted by depositing $CVNT into the settlement program and burned at the point of consumption, with each burn bound to a receipt batch whose Merkle root is anchored on Solana.
This makes agent activity accountable and auditable: an operator can reconcile precisely what an agent did and what it cost, down to the receipt. $CVNT utility is functional — it is the unit agents pay in to consume compute, memory, tools, and paid external services through the layer.
Licensed Apache-2.0, with public documentation and a live sandbox at opencovenant.org.
Risk Report
Cryptocurrency carries risk. Every project listed on this site, regardless of its score, operates in a volatile and largely unregulated market. Tokens can lose most or all of their value. A high score reflects lower structural risk based on the data we can measure. It does not mean a project is safe, that the price will hold, or that losses are unlikely. Always research independently and never commit money you cannot afford to lose.
