The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data in the form of academic papers, web articles, and technical reports for summarization and citation extraction.
Ingestion points: User-provided text and documents are processed via the /research:summarize and /research:cite commands.
Boundary markers: The skill does not currently define specific boundary markers or instructions to isolate the source text from the agent's control logic.
Capability inventory: The skill does not possess dangerous capabilities such as network exfiltration, arbitrary command execution, or sensitive file system writes. The included scripts are restricted to string manipulation using the Python standard library.
Sanitization: No explicit sanitization of input text is implemented, but the lack of exploitable capabilities mitigates this surface.
[SAFE]: The Python scripts (scripts/extract_citations.py and scripts/format_summary.py) utilize standard libraries for regular expression processing and template rendering. No remote code execution, dynamic code loading, or obfuscation techniques were identified.

research-summarizer