The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/search.sh script takes user input for the search query and passes it to the grep command without sanitization. This allows for potential argument injection where specially crafted queries could modify the command's execution flags.
[EXTERNAL_DOWNLOADS]: The skill's documentation and metadata reference an external script (~/fetch_drli.mjs) used for updating the local article database from drli.beaucare.org. Since the code for this script is not included in the skill package, its behavior and safety cannot be verified.
[PROMPT_INJECTION]: The skill ingests and processes article content from an external website, which creates a vulnerability to indirect prompt injection if those articles contain malicious instructions.
Ingestion points: Article and index JSON files stored in the home directory (~/drli_articles/ and ~/drli_articles_index.json).
Boundary markers: None identified; the agent reads and prints article contents directly without delimiters.
Capability inventory: File system access via grep and cat, plus Python-based data extraction and a referenced maintenance script.
Sanitization: No sanitization or validation of the article text is performed before it is processed by the agent.

binlee-articles