The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the postplus CLI tool to perform all operations, including managing connections, listing tools, and executing messaging tasks.
Evidence: Multiple instances of postplus channels tools run, postplus media-file upload, and postplus channels connect commands are used throughout SKILL.md and the reference files to interact with external APIs and local files.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it ingests untrusted data from external messaging services that could be used to influence the agent's behavior.
Ingestion points: Message history and content are retrieved via tools such as DISCORDBOT_GET_MESSAGE (references/discord.md), SLACK_FETCH_MESSAGE_THREAD_FROM_A_CONVERSATION (references/slack.md), and WHATSAPP_GET_MESSAGE_HISTORY (references/whatsapp-business.md).
Boundary markers: The instructions require the agent to "Resolve five facts before any send" and verify parent message context, serving as a procedural boundary, although no technical escaping is specified.
Capability inventory: The agent has extensive write permissions, including sending messages (SLACK_SEND_MESSAGE, DISCORDBOT_CREATE_MESSAGE, WHATSAPP_SEND_MESSAGE) and uploading local files to the platform (postplus media-file upload).
Sanitization: No explicit instructions for sanitizing or escaping the ingested content are provided before the agent uses it to draft or send messages.

channel-messaging