The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The provided code examples demonstrate patterns where untrusted data from Slack users is interpolated directly into messages sent by the bot. While benign in a standard bot, this creates a vulnerability surface for indirect prompt injection if the bot's outputs are later processed by an LLM-driven agent.
Ingestion points: command.text from the /deploy slash command and selected_option.value from the deploy_modal in SKILL.md.
Boundary markers: No delimiters or instructions are used to separate user-provided content from the bot's own text.
Capability inventory: The skill uses client.chat.postMessage, respond, and say to write to Slack channels.
Sanitization: No escaping or validation is performed on the user-supplied strings before they are posted to the channel.
[EXTERNAL_DOWNLOADS]: The skill instructs users to install @slack/bolt via npm. This is the official, well-known SDK for building Slack applications.

slack