The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection by processing external sequence databases and search result files without comprehensive sanitization of string identifiers.
Ingestion points: The scripts/parse_blast_tab.py script reads and parses sequence search results from user-supplied local files.
Boundary markers: The skill does not implement specific delimiters or explicit instructions to the agent to treat search result content as untrusted data during the parsing workflow.
Capability inventory: The skill is authorized to use Bash for command execution (restricted to sequence search tools) and Read/Write tools for file management.
Sanitization: While the parsing script coerces numeric metrics (e-values, bitscores) to float types for filtering, it does not sanitize sequence IDs or descriptions before they are returned to the agent context.
[EXTERNAL_DOWNLOADS]: The skill references and relies on binaries and documentation from well-known bioinformatics organizations and repositories.
Fetches documentation and provides command references for NCBI BLAST+ tools.
References the official DIAMOND sequence aligner from its public GitHub repository.
Notes compatibility and installation via the Bioconda package registry.

alterlab-blast