The Agent Skills Directory

Proactive Data Redaction: The skill incorporates a safety gate using grep with regular expressions to scan draft reports for passwords, API keys, tokens, and GUIDs. This ensures that sensitive environment details are identified for redaction before a user reviews the final draft.
Human-in-the-Loop Verification: The skill is designed to generate a URL for the user to open and submit manually. This approach ensures the user maintains full control over the reporting process and can perform a final review of all content before it is posted to a public repository.
Secure Data Ingestion: When capturing user feedback, the skill uses the cat <<'EOF' pattern. The use of single quotes around the EOF delimiter prevents the shell from performing variable expansion or command substitution on the input, effectively neutralizing simple injection attempts during the drafting phase.
Trusted Resource Integration: All network operations, including checking repository status and fetching issue templates, target official Microsoft domains (aka.ms) and the vendor's official GitHub organization (microsoft/microsoft-sql). The use of standard tools like gh and curl for these tasks aligns with best practices for interacting with established infrastructure.

skill-feedback