The Agent Skills Directory

[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by instructing the agent to observe live message traffic and logs from the Ably platform. * Ingestion points: The agent is directed to use ably subscribe, ably presence subscribe, and ably logs to monitor real-time data from Ably channels (found in SKILL.md). * Boundary markers: The instructions lack specific delimiters or warnings to ignore commands that may be embedded within the messages being observed. * Capability inventory: The agent can use the Ably CLI to publish messages, manage API keys, and configure integrations (found in SKILL.md). * Sanitization: There is no mention of sanitization or validation of the data received from these live streams.
[DATA_EXFILTRATION]: The skill provides guidance for the agent to access sensitive credentials. It suggests searching for the ABLY_API_KEY or ABLY_TOKEN within 'the project's environment files' (e.g., .env) to facilitate CLI authentication (found in SKILL.md).
[COMMAND_EXECUTION]: The skill's primary functionality is based on executing various ably CLI commands to interact with the Ably service (found in SKILL.md).
[EXTERNAL_DOWNLOADS]: The documentation mentions installing the @ably/cli package from the npm registry. This is an official tool provided by the vendor (found in SKILL.md).

debugging-with-ably-cli