The Agent Skills Directory

[SAFE]: The skill implements a legitimate research tool. It uses the openai and requests libraries to communicate with api.parallel.ai and openrouter.ai. API keys are managed through standard environment variables (PARALLEL_API_KEY, OPENROUTER_API_KEY), which is a recommended security practice.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to api.parallel.ai and openrouter.ai (OpenRouter) to fetch research data. These are the primary service endpoints for the skill's functionality.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute its own Python scripts (lookup.py, research_lookup.py). These scripts are provided within the skill and do not contain arbitrary or dangerous command execution logic.
[PROMPT_INJECTION]: Analysis of SKILL.md and README.md shows no attempts to override agent safety guidelines or bypass instructions. The instructions are focused on the task of research routing and citation quality.
[DATA_EXFILTRATION]: No sensitive local data access (like .ssh or .aws credentials) was detected. The only data sent over the network are the user's research queries, directed to the official backend APIs.
[SAFE]: The skill mandates saving research results to a local sources/ folder for reproducibility and auditability, which is a beneficial practice for scientific research workflows.

research-lookup