The Agent Skills Directory

[SAFE]: The skill utilizes industry-standard and well-known libraries (@solana/web3.js, tweetnacl, and bs58) for all cryptographic and blockchain operations.
[SAFE]: Security best practices are implemented through configurable transaction limits (MAX_SOL_PER_TX), a mandatory dry-run simulation mode by default, and human-in-the-loop requirements for transactions above a specified threshold.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the Solana blockchain, such as transaction history and token metadata, which could contain malicious instructions embedded by external actors.
Ingestion points: Data enters the agent context through the getTransactions and getTokenAccounts functions in scripts/solana.js.
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded content when processing blockchain data.
Capability inventory: The sendSol function in scripts/solana.js grants the agent the capability to move funds.
Sanitization: No sanitization or filtering of blockchain-derived strings is performed before they are presented to the agent.

solana-connect