The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructions manage sensitive Hive private keys (Posting and Active keys). It directs users to store these in environment variables or a local configuration file at ~/.hive-tx-cli/config.json.
[EXTERNAL_DOWNLOADS]: The skill configuration initiates the installation of the @peakd/hive-tx-cli package from the npm registry during setup.
[DATA_EXFILTRATION]: The hive upload command functionality transmits local image files to external hosting providers such as https://images.ecency.com.
[COMMAND_EXECUTION]: The skill executes the hive binary with various subcommands and arguments, some of which may be derived from external data or user input.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from an external source (the Hive blockchain) and has high-impact capabilities (financial transfers).
Ingestion points: Data is fetched from the blockchain via hive content, hive replies, hive feed, and hive account commands (as seen in SKILL.md).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore instructions' warnings when the agent processes the retrieved blockchain content.
Capability inventory: The skill can perform hive transfer, hive vote, hive publish, hive claim, and hive delegate operations.
Sanitization: There is no mention of sanitizing or validating the content retrieved from the blockchain before it is processed by the agent.

hive