The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to install the core CLI tool by piping a remote shell script directly into the system shell: curl -fsSL https://raw.githubusercontent.com/stablyai/agent-slack/main/install.sh | sh. This pattern allows arbitrary code execution from a third-party source without prior verification.
[EXTERNAL_DOWNLOADS]: In addition to the shell script, the skill promotes downloading code via npm i -g agent-slack and nix run github:stablyai/agent-slack from the stablyai organization, which is not the skill's author.
[CREDENTIALS_UNSAFE]: The tool includes built-in commands designed to extract sensitive authentication data from the host environment:
agent-slack auth import-chrome and agent-slack auth import-firefox harvest session cookies and credentials from local browser profiles.
agent-slack auth import-desktop extracts authentication data from the local Slack Desktop application storage.
[DATA_EXFILTRATION]: The skill documentation indicates that the message get, message list, and search commands automatically download all attachments (files, images, and snippets) to the local filesystem (~/.agent-slack/tmp/downloads/) whenever they are encountered. This behavior can lead to the silent collection of external data onto the agent's host machine without explicit user consent for each file.

agent-slack