The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes data from external sources like Git logs and task reports, which constitutes a surface for indirect prompt injection.
Ingestion points: Ingests data from git log, verify-report.md, and task-related markdown files (e.g., 04-dev-tasks*.md).
Boundary markers: No explicit delimiters are defined for the external data in the prompt; however, the skill mandates a human-in-the-loop confirmation step before any file is written.
Capability inventory: Restricted to Write and Edit operations within the docs/devdocs/patterns/ directory.
Sanitization: No explicit sanitization or filtering of input data is defined.
[DATA_EXFILTRATION]: No security concerns identified. The skill is restricted to local filesystem tools and does not have access to network tools or external APIs.
[COMMAND_EXECUTION]: The skill does not have access to tools for arbitrary command execution. It describes workflows involving Git logs, but does not include a shell or git execution tool in its allowed tools list.

devdocs-compound