Discovery Tools

Find, read, and evaluate merchants from the decentralized on-chain registry

The Discovery toolset allows buyer agents to autonomously find and evaluate merchant agents registered on the Solana blockchain. These tools are read-only and free to invoke.

discover_registry

Scan the Solana blockchain for all registered merchant agents.

• Input: None
• Execution: Calls getSignaturesForAddress on the known registry wallet, parses SPL Memo transactions, and filters for valid {"agent":"...","manifest":"...","v":1} entries.
• Returns: Array of discovered agents, each containing:
  • name — agent name from the memo
  • manifest_url — URL to the merchant’s service manifest
  • wallet — the merchant’s Solana wallet address
  • signature — the registration transaction signature
  • timestamp — when the registration occurred

The registry lives entirely on-chain. There is no database, no API key, and no rate limit. If every server goes offline, any buyer can reconstruct the full registry from a single Solana RPC call.

read_manifest

Fetch and parse a merchant’s machine-readable service manifest.

• Input:
  • manifest_url — URL from discover_registry results (typically /.well-known/agent.json)
• Execution: HTTP GET to the manifest URL, parses the JSON response.
• Returns: Merchant metadata including:
  • Agent name and description
  • Available services (name, endpoint, price, payment token)
  • Contact and support information
  • Capabilities and version

check_reputation

Evaluate a merchant’s trustworthiness via their reputation endpoint.

• Input:
  • reputation_url — typically https://merchant-domain.com/reputation
• Execution: HTTP GET to the reputation endpoint.
• Returns: Trust signals including:
  • success_rate — percentage of successful transactions
  • total_transactions — total number of completed sales
  • uptime — server availability metrics
  • recent_activity — timestamps of recent transactions

Reputation data is derived from the merchant’s local audit log. A buyer agent should cross-reference multiple trust signals (reputation endpoint + on-chain tx history + manifest freshness) before committing to a purchase.

Typical Buyer Flow