The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dynamically resolves the file paths for utility scripts like research_wiki.py and DEEPXIV_FETCHER using shell logic, then executes them via python3. Executing code from paths determined at runtime is a risk factor.
[EXTERNAL_DOWNLOADS]: Fetches research data from the ArXiv API and AlphaXiv overview pages.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection.
Ingestion points: External research paper summaries are retrieved via WebFetch and curl from remote sources.
Boundary markers: The skill lacks explicit delimiters or instructions for the agent to ignore potentially malicious directions embedded within the fetched text.
Capability inventory: The agent can modify local files through the Edit tool and execute shell commands via Bash, which could be leveraged if malicious instructions were present in a processed research abstract.
Sanitization: There is no documented validation or sanitization of the fetched external text before it is processed by the agent to generate wiki updates.

wiki-enrich