The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data in the form of Slack messages. It includes robust instructions to mitigate injection risks, directing the agent to categorize message intent rather than repeating content verbatim, and explicitly prohibiting the interpretation of message content as instructions.
[COMMAND_EXECUTION]: The skill utilizes several Slack API tools, including slack_search_public_and_private, slack_read_channel, and slack_send_message. These tools are employed correctly and safely within the context of the skill's intended purpose of managing Slack communications.
[DATA_EXFILTRATION]: The skill handles sensitive communication data such as DMs and private channel mentions. However, this data is used exclusively to provide a summary to the authenticated user within the agent's interface. No unauthorized data transmission to external or third-party domains was observed.
[PROMPT_INJECTION]: A static analysis warning was triggered due to phrases like 'ignore previous instructions'. Upon manual review, these phrases are part of a defensive instruction set intended to help the agent recognize and neutralize injection attempts in message content, rather than an attempt to override the agent's core safety protocols.

p-slack-triage