The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions fetch requirements files (install-requirements.txt, requirements.txt) and a pre-compiled Python wheel (tuplatform_connect) from https://connect.aiscientist.tools and GitHub repositories.
[REMOTE_CODE_EXECUTION]: The skill directs the user to install a Python wheel directly from a remote URL (https://connect.aiscientist.tools/downloads/tuplatform_connect-0.3.0-py3-none-any.whl). Although the instruction includes a SHA-256 hash for integrity verification, this pattern involves executing binary code from a non-standard third-party source.
[COMMAND_EXECUTION]: The setup guide includes numerous shell commands for environment creation with uv, dependency management with pip, and service orchestration using the tu (ToolUniverse) CLI tool.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an MCP (Model Context Protocol) server designed to process biological data files, creating a surface for indirect prompt injection.
Ingestion points: The run_slingshot_trajectory operation accepts an adata_path parameter (SKILL.md), which is used to load external data files into the agent's context.
Boundary markers: No explicit boundary markers or instructions to ignore embedded content within data files are provided in the implementation examples.
Capability inventory: The setup grants the ability to execute shell commands via the tu CLI and run local Python scripts that interface with the ToolUniverse platform.
Sanitization: There is no evidence of input validation or sanitization for the data files processed by the tool.

setup-slingshot-remote-tool