The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the anndata package and its ecosystem (such as scanpy and muon) using the uv pip package manager. These are well-established, standard libraries in the bioinformatics and scientific computing communities.
[INDIRECT_PROMPT_INJECTION]: The skill provides the ability to ingest and process data from various external file formats and remote sources, which represents a potential attack surface for indirect prompt injection if those files contain adversarial strings.
Ingestion points: The skill reads data from external sources via ad.read_h5ad, ad.read_csv, ad.read_zarr, ad.read_loom, and ad.read_10x_h5 as documented in SKILL.md and references/io_operations.md.
Boundary markers: No specific delimiters or boundary markers are established to prevent the AI from interpreting data content (metadata strings) as instructions.
Capability inventory: The skill possesses the capability to write files (adata.write_h5ad) and perform network-based data retrieval using fsspec and urllib.
Sanitization: There is no evidence of internal sanitization or validation mechanisms to filter potentially malicious instruction strings within the observation or variable metadata of the data matrices.

anndata