The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches specification data from 'https://agentskills.io/specification' to ensure compliance with the current agent skill standards.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and acts upon external data sources such as the fetched specification and the user's repository contents. * Ingestion points: Specification content from 'https://agentskills.io/specification' and repository files inspected during the 'research' phase. * Boundary markers: None identified for delimiting external specification data from instructions. * Capability inventory: Subprocess execution via browser automation, web search, and file writing operations. * Sanitization: The skill contains explicit instructions to avoid accessing credentials or private data and to request sanitized examples from the user for sensitive inputs.

skill-writer