The Agent Skills Directory

[SAFE]: The skill is designed for knowledge management and operates primarily on a user-specified local directory. The provided scripts for initialization, linting, and statistics are transparent and perform only their documented functions without hidden or malicious logic.
[EXTERNAL_DOWNLOADS]: The arXiv extension fetches academic paper metadata from export.arxiv.org. This is a well-known, reputable service used as intended for research ingestion. The GitHub extension uses the official gh CLI or GitHub API, following standard integration patterns.
[COMMAND_EXECUTION]: The skill uses PowerShell commands and Python scripts to automate file management and data processing. These operations are scoped to the wiki directory and are necessary for the skill's primary functions, such as creating directory structures and verifying link integrity.
[INDIRECT_PROMPT_INJECTION]: By design, the skill ingests content from external sources into the agent's context. While this creates a surface for indirect prompt injection, the skill includes mitigation strategies such as mandatory human-in-the-loop review of takeaways and automated linting to detect inconsistencies in the compiled knowledge base.
[CREDENTIALS_SAFE]: The skill demonstrates secure handling of sensitive data by explicitly instructing users to use environment variables for GitHub authentication and warning against hardcoding API tokens in wiki files.

llm-wiki