The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface in the product feedback synthesis component.
Ingestion points: The agent is instructed to use WebFetch and Read to gather untrusted content from app store reviews, Reddit, and social media platforms in references/domains/misc/product-feedback-synthesizer.md.
Boundary markers: No explicit delimiters or instructions are provided to differentiate between system prompts and external data.
Capability inventory: The agent has access to Read, Write, Grep, WebFetch, and MultiEdit tools in references/domains/misc/product-feedback-synthesizer.md.
Sanitization: There is no defined process for validating or filtering ingested content.
[COMMAND_EXECUTION]: The skill uses local scripts for task automation and repository management.
Evidence: Deterministic Python scripts market-sizing.py and user-story-template.py are provided to generate calculation results and Markdown formatting.
Evidence: Maintenance shell scripts (e.g., scripts/add-a-skill.sh, scripts/test-a-skill.sh) are referenced in references/domains/product-management/skill-authoring-workflow/SKILL.md for administrative tasks.
[EXTERNAL_DOWNLOADS]: Certain components are designed to fetch data from external domains for research purposes.
Evidence: The trend-researcher and feedback-synthesizer agents are configured to retrieve information from public platforms like TikTok, YouTube, and various application marketplaces to identify market trends.

product-specialist