The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses the GitHub CLI (gh repo clone) to download external repository content for analysis. These operations target a well-known service (GitHub) and are central to the skill's stated purpose of reviewing external projects.
[COMMAND_EXECUTION]: The skill performs shell-based analysis using grep and git log to discover process files and historical run data within the cloned repositories. These are standard diagnostic operations for code analysis tools.
[PROMPT_INJECTION]: (Category 8: Indirect Prompt Injection) This skill has an inherent attack surface because it ingests untrusted data from external repositories.
Ingestion points: External code files and JSON run logs are loaded into the agent's context during Phases 2 and 4.
Boundary markers: The skill includes explicit instructions to perform a 'read-only analysis' and a strict safety warning: 'Never execute the external process'.
Capability inventory: The skill possesses file-write capabilities (to docs/retrospectives/) and shell execution capabilities for repository management and searching.
Sanitization: There is no explicit logic described for sanitizing input data before processing, though the focus is on structured retrospective reporting.

retrospect-external-babysitter-run