The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill manages authentication tokens in the ~/.sendblue/credentials.json file. The documentation includes the sendblue whoami command, which is noted to expose these credentials to the terminal output. If executed by an agent, this would place sensitive API keys into the session history and context.- [INDIRECT_PROMPT_INJECTION]: The skill provides functionality to read inbound SMS and iMessages via the sendblue messages --inbound command. This creates a risk where malicious instructions sent to the user's Sendblue number could influence the agent's behavior when it processes those messages.
Ingestion points: Inbound message content retrieved via sendblue messages.
Boundary markers: Absent. The instructions do not recommend using delimiters or specific 'ignore' instructions for inbound message text.
Capability inventory: The skill possesses shell execution capabilities (via npm and the sendblue CLI) and file system access.
Sanitization: Absent. No evidence of content filtering or escaping for retrieved message text.- [EXTERNAL_DOWNLOADS]: The skill instructs the installation and use of the @sendblue/cli npm package. While this is the official tool for the service, it involves downloading code from a public package registry.- [COMMAND_EXECUTION]: The skill's operational model is built entirely around shell command execution, providing multiple points for interaction with the underlying system terminal.

sendblue-cli