The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to orchestrate a pipeline of standard system utilities, including git for history analysis, gh for GitHub API interaction, and sqlite3 for data persistence. These operations are scoped to the repositories identified by the user and follow authenticated session protocols.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content such as commit messages and PR titles which represents a potential vulnerability surface. 1. Ingestion points: Data enters via dxm-ingest-git.sh and dxm-ingest-github.sh. 2. Boundary markers: Present. Output templates in templates/dashboard.html include explicit banners and proxy disclosures. 3. Capability inventory: Use of git, sqlite3, awk, jq, and gh across all logic files. 4. Sanitization: Present. The system escapes data before output and includes an automated privacy check in dxm-trends.sh to prevent PII exposure.
[DYNAMIC_EXECUTION]: The scripts generate and execute local SQL statements at runtime to handle bulk data operations. This is a standard implementation for the skill's data-heavy workflow and does not involve untrusted code execution or remote script loading.

dx-metrics