The Agent Skills Directory

[METADATA_POISONING]: The YAML frontmatter in SKILL.md identifies the author as 'K-Dense Inc.', which differs from the platform-provided author identity 'leonchaox'. While not inherently malicious, this discrepancy represents a metadata inconsistency.
[INDIRECT_PROMPT_INJECTION]: The skill ingests biochemical data (kinetic parameters and reaction information) from the external BRENDA SOAP API, which is subsequently processed into reports and visualizations. This creates a potential surface for indirect prompt injection if the external database were to contain malicious instructions.
Ingestion points: Data enters the system via the get_km_values and get_reactions functions in scripts/brenda_queries.py.
Boundary markers: The skill does not utilize specific delimiters or instructions to prevent the agent from interpreting ingested data as commands.
Capability inventory: The skill has capabilities to write text reports, export structured data files (CSV, JSON), and generate PNG visualizations.
Sanitization: The skill relies on regular expressions for data extraction but does not sanitize the contents of database commentary fields for potential injection strings.
[DYNAMIC_EXECUTION]: The utility script scripts/brenda_queries.py performs aggressive path manipulation by adding a directory four levels above the script's location to the sys.path. This dynamic path calculation is used to import the brenda_client module but represents a non-standard loading pattern that could potentially allow the skill to access or import modules from outside its own directory structure.

brenda-database