The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted document content (academic papers, articles, reports), which naturally creates a surface for indirect prompt injection.
Ingestion points: Document text ingested for analysis via the /research:summarize, /research:compare, and /research:cite workflows.
Boundary markers: The skill instructions do not explicitly implement delimiters or ignore-instructions for the processed source text.
Capability inventory: The skill utilizes local Python scripts (scripts/extract_citations.py and scripts/format_summary.py) to parse and format document data.
Sanitization: No explicit filtering or instruction-scrubbing is performed in the provided scripts before the text is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill's installation documentation includes references to external repositories and registries for acquisition.
Evidence: Mentions of github.com/alirezarezvani/claude-skills.git and the clawhub registry. These references are documented as part of the standard installation process and are attributed to the skill's author.

research-summarizer