The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from the Solana blockchain, which is an external and untrusted source of information.
- Ingestion points: Data is ingested through tools like parseTransactions, getTransactionHistory, and getAssetsByOwner, which retrieve potentially attacker-influenced content such as transaction descriptions and NFT metadata (SKILL.md, references/enhanced-transactions.md, references/das.md).
- Boundary markers: The instructions do not specify delimitation strategies or warnings for the agent to ignore instructions that might be embedded in the retrieved blockchain content.
- Capability inventory: The agent has the ability to execute on-chain operations (transferSol, transferToken), manage credentials (setHeliusApiKey), and configure alerting systems (createWebhook).
- Sanitization: The skill utilizes the Helius Enhanced Transactions API which provides parsed, human-readable summaries, reducing the risk of accidental command execution from raw instruction data, though natural language risks remain.
[EXTERNAL_DOWNLOADS]: The skill references and provides installation instructions for several external tools and libraries.
- Evidence: Instructions to install helius-mcp and helius-cli using npx from the NPM registry (SKILL.md, references/onboarding.md).
- Context: These are documented installation paths for official Helius tools from a standard package registry.

helius