The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is designed to handle Solana keypair files (e.g., ~/.config/solana/id.json), which contain sensitive, unencrypted private keys. This establishes a high-value target for potential credential harvesting.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to interact with the filesystem, reading and validating the content of these sensitive credential files.
[DATA_EXFILTRATION]: The instructions require the agent to read the content of private key files to extract the public address. This process brings the raw private key into the agent's context, creating a risk where the secret could be leaked in the model's output or exfiltrated if the session is compromised.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes external data from local files without established boundary markers.
Ingestion points: Reads local files specified via arguments or the SOLANA_KEYPAIR_PATH environment variable.
Boundary markers: None identified in the skill instructions.
Capability inventory: Includes Bash for file reading and system interaction.
Sanitization: No sanitization of the file content is performed prior to processing by the agent.

solana-keypair