The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the ingestion of untrusted external data.
Ingestion points: The skill extensively uses web_fetch to retrieve articles and data from across the internet during the market research phase (documented in SKILL.md).
Boundary markers: The instructions lack specific boundary markers or instructions for the agent to treat fetched data as untrusted, which could allow malicious content in fetched articles to influence the agent's logic.
Capability inventory: The skill possesses the ability to write local JSON files and execute a Python script (scripts/market_analyzer.py) to process that data.
Sanitization: No explicit sanitization, filtering, or validation logic is defined for the content retrieved from external sources before it is incorporated into the analysis or report.
[COMMAND_EXECUTION]: The skill includes instructions to execute a bundled Python script for data processing.
Evidence: SKILL.md directs the agent to execute python scripts/market_analyzer.py analysis_data.json. While the bundled script uses safe practices like json.load() and standard arithmetic, this represents an active command execution capability that relies on the agent correctly formatting the input data.

startup-validator