The Agent Skills Directory

[SAFE]: The skill instructions focus on standard development workflow practices such as version policy enforcement, security risk assessment, and lockfile integrity. The instructions explicitly direct the agent to prioritize safety, smallest practical changes, and phased risk control.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze third-party dependency data, which presents a surface for indirect prompt injection.
Ingestion points: The instructions specify analyzing package graphs, lockfiles, and transitive dependencies from the workspace (SKILL.md).
Boundary markers: No specific boundary markers or instruction-ignore delimiters are defined in the instructions.
Capability inventory: The skill is granted workspace-write access in the frontmatter metadata, allowing it to modify the file system.
Sanitization: While the instructions emphasize validation (e.g., 'validate one normal path, one failure path'), they do not specify technical sanitization or filtering of external package data.

dependency-manager