The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and process text messages from external agents using the pilotctl inbox and pilotctl listen commands. This creates an indirect prompt injection surface where a malicious sender could embed instructions to manipulate the receiving agent's behavior.
Ingestion points: Incoming message data retrieved via pilotctl inbox and pilotctl --json listen 7 as described in SKILL.md.
Boundary markers: None. There are no instructions or delimiters provided to help the agent distinguish between its own system prompt and the untrusted data received from the network.
Capability inventory: The skill explicitly allows the Bash tool and demonstrates using it for command substitution and piping to utilities like jq.
Sanitization: The provided workflow and command examples do not include any validation, escaping, or sanitization of the received message content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill's core functionality relies on executing shell commands via the Bash tool to interact with the pilotctl binary. While these commands are specific to the protocol's intended use, they provide the agent with active control over a local network-interfacing utility.

pilot-chat