The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external RCSB APIs, which is then processed by the agent, creating a surface for indirect prompt injection.
Ingestion points: API responses from search_pdb.py, fetch_pdb_metadata.py, and fetch_schema.py are stored in local files and read by the agent.
Capability inventory: The skill has the ability to execute internal scripts via uv run and access network resources on rcsb.org domains.
Boundary markers: The instructions lack explicit markers or warnings to the agent to treat the external data as untrusted.
Sanitization: No sanitization or filtering of the retrieved data is performed before agent processing.
[EXTERNAL_DOWNLOADS]: The skill fetches scientific data, metadata, and configuration schemas from official RCSB Protein Data Bank infrastructure (rcsb.org, search.rcsb.org, data.rcsb.org, files-beta.rcsb.org). These downloads are essential for the skill's primary function and target well-known scientific sources.
[COMMAND_EXECUTION]: Utility scripts provided with the skill are executed using uv run to perform data retrieval and processing tasks. This is a standard practice for modular AI agent skills.

pdb-database