The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides instructions and code for setting up HTTP servers that process requests from external AI agents, creating an inherent surface for indirect prompt injection.
Ingestion points: The payment server implementation (e.g., in mpp.md and x402.md) creates Express routes that receive input from untrusted network clients.
Boundary markers: The technical snippets focus on payment verification and do not explicitly define delimiters or instruction-ignoring wrappers for data processed at the application level.
Capability inventory: The skill possesses network capabilities (e.g., mppx.fetch, Horizon SDK for ledger submission) and local filesystem access used during the initial environment configuration.
Sanitization: While the underlying payment libraries validate protocol credentials, the skill does not detail application-level sanitization for the specific content delivered by the API after payment is confirmed.
[EXTERNAL_DOWNLOADS]: The skill facilitates the acquisition of resources from well-known and trusted infrastructure within the Stellar ecosystem.
Fetches testnet funds from the official Stellar Friendbot service (friendbot.stellar.org).
Guides users to obtain testnet USDC from the Circle faucet service.
Recommends the installation of verified libraries from the NPM registry, including official @stellar packages and OpenZeppelin-facilitated payment tools.

agentic-payments