The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill includes commands to import Slack authentication tokens from local application data and browser profiles, such as slack auth import-desktop, slack auth import-chrome, and slack auth import-firefox (documented in references/commands.md). While functional for setup, this involves the extraction of sensitive credentials from other software on the system.\n- [EXTERNAL_DOWNLOADS]: As part of retrieving messages or performing searches, the CLI tool automatically downloads attachments to local directories like ~/.agent-slack/tmp/downloads/. This results in external, potentially untrusted files being written to the local filesystem (documented in references/output.md).\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted content from Slack. \n
Ingestion points: The skill fetches data via slack message get, slack message list, and slack search (references/commands.md).\n
Boundary markers: There are no specific delimiters or instructions to the agent to disregard commands within the fetched message content.\n
Capability inventory: The agent has the authority to perform impactful actions like sending, editing, or deleting messages, and inviting users to channels (SKILL.md).\n
Sanitization: No evidence of sanitization or content validation is provided in the documentation.\n- [COMMAND_EXECUTION]: The skill performs sensitive workspace operations, such as managing user invites and channel configurations, by executing the slack CLI tool with high-level permissions.

slack