The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions in references/raptor_setup.md to clone a third-party repository (https://github.com/solanatracker/raptor-binary) containing an external binary.
[COMMAND_EXECUTION]: The setup guide instructs users to grant execution permissions (chmod +x raptor) and run the external binary (./raptor) locally. Executing opaque binaries from unverified sources poses a significant security risk.
[CREDENTIALS_UNSAFE]: The skill and its associated scripts (scripts/wallet_pnl.py, references/raptor_setup.md) handle highly sensitive credentials, including a Solana PRIVATE_KEY for transaction signing and a SOLANATRACKER_API_KEY for API access.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via attacker-controlled data. Ingestion points: Token metadata (names, symbols, risk descriptions) fetched from the SolanaTracker API in scripts/token_analysis.py. Boundary markers: Absent; data is interpolated directly into report outputs. Capability inventory: Scripts output fetched data to the console where it is processed by the agent. Sanitization: None; the skill does not filter or sanitize external token strings before display.

solanatracker-api