The Agent Skills Directory

[SAFE]: The skill implementation follows security best practices by utilizing only the Python standard library, which avoids risks associated with third-party dependency vulnerabilities or supply chain attacks.
[SAFE]: Authentication credentials, including $YUQUE_COOKIES and $YUQUE_TOKEN, are handled with significant care. The YuqueClient class in scripts/yuque.py ensures that cookies are only sent to .yuque.com subdomains. Furthermore, a custom NoRedirectHandler is implemented to explicitly block HTTP redirects, ensuring that sensitive headers like x-csrf-token or session cookies are never leaked to external or untrusted domains.
[SAFE]: The skill incorporates a robust write-gating mechanism for destructive actions. Commands such as create, update, and delete default to a 'dry-run' mode and require the --confirm flag as the final argument to execute. This prevents accidental data modification and mitigates the risk of command injection through document content.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from Yuque, which introduces a vulnerability surface for indirect prompt injection.
Ingestion points: The doc and docs methods in scripts/yuque.py retrieve document titles and markdown content from Yuque's API.
Boundary markers: None are present; the raw content is returned to the agent context.
Capability inventory: The skill allows the use of Bash and maintains network access to yuque.com for document management.
Sanitization: The script outputs data in JSON format but does not filter or sanitize the markdown content retrieved from the remote source.

yuque