The Agent Skills Directory

[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests untrusted data from web search results (via Tavily or search_web) and academic databases (OpenAlex, Wikidata). This external content is interpolated into workflow prompts and generated scripts without robust boundary markers or explicit sanitization, potentially allowing third-party content to influence agent behavior.- [REMOTE_CODE_EXECUTION]: The skill dynamically generates and likely executes a Python script (generate_techmap.py) that uses the graphviz library. Since the script's content is partially derived from external data (e.g., person names, company descriptions), there is a risk of code injection if retrieved content contains malicious Python code designed to escape string literals.- [EXTERNAL_DOWNLOADS]: The skill communicates with external services, including the well-known Wikidata SPARQL endpoint (query.wikidata.org) and academic databases. These network operations are documented and necessary for the skill's primary research purpose, representing a typical use of established technological resources.

technology-mapping