The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to download and install a Python wheel file from a remote domain (https://connect.aiscientist.tools/downloads/tuplatform_connect-0.3.0-py3-none-any.whl). While the instruction includes a SHA-256 hash for integrity verification (sha256=3fad5eee5ecf7887a693d93ccd1aa112dc0955617a885d1fc3daded0030f9ae0), the source is not a standard package registry or a recognized trusted vendor repository.
[DYNAMIC_EXECUTION]: The skill uses shell heredocs to dynamically generate and execute Python scripts (e.g., using `python
<<'PY'`) to perform tool discovery and verification of the local MCP endpoint.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data files via the adata_path parameter in the run_scrublet_doublets tool, which creates a vulnerability to instructions embedded within untrusted data.
Ingestion points: The adata_path argument provided to the run_scrublet_doublets operation.
Boundary markers: None identified. There are no instructions for the agent to ignore or delimit instructions found within the input files.
Capability inventory: The skill can execute Python scripts via heredocs, install packages, and invoke MCP tools.
Sanitization: No sanitization or validation of the input file contents (e.g., H5AD count matrices) is described to prevent instruction injection.

setup-scrublet-remote-tool