The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill utilizes the readRDS function in scripts/utils.R to load interaction databases from disk. readRDS is a deserialization function in the R programming language that is considered insecure because it can execute arbitrary code contained within a maliciously crafted .rds file upon loading. This vulnerability is exposed because the skill allows users to specify a custom database file path using the --db_path parameter in scripts/main.R.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted gene list data from user-supplied strings or files via the --gene and --gene_file arguments in scripts/main.R. While the parse_gene_input function in scripts/utils.R performs basic formatting and filters for unique symbols, it does not implement robust sanitization or boundary markers to isolate user data from the agent context. This creates a surface for indirect prompt injection, although the risk is low given the skill's specific analytical focus and limited capabilities.

tf-target-gene-regulatory-network