The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to install and execute the yoktez-mcp tool directly from a third-party GitHub repository using the uvx command (uvx --from git+https://github.com/saidsurucu/yoktez-mcp yoktez-mcp). This involves fetching and running code from an external repository not affiliated with the skill author.
[EXTERNAL_DOWNLOADS]: The skill relies on several external services and endpoints to function, including a hosted connector at yoktezmcp.fastmcp.app and a community mirror for metadata at tezara.org.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from the National Thesis Center database, creating a vulnerability surface for indirect prompt injection.
Ingestion points: Thesis metadata including titles, keywords, and abstracts retrieved via the search_yok_tez_detailed tool.
Boundary markers: The skill instructions do not define clear delimiters or escaping mechanisms to separate retrieved thesis content from agent instructions.
Capability inventory: The agent has access to Bash, Write, and Edit tools while processing the untrusted data.
Sanitization: No sanitization or filtering logic is provided to validate or clean the content of the thesis records before they are interpolated into the prompt.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute a local Python helper script (scripts/yok_tez_query.py) to perform query normalization and format citations.

alterlab-yok-tez