The Agent Skills Directory

[SAFE]: The skill provides a structured framework for cross-session memory management, using explicit human-in-the-loop approval for context modifications through the apply_context_patch tool.- [PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection by design, as it processes and retrieves stored user and session data.
Ingestion points: The read_context, search_context_topics, and append_capture tools in SKILL.md are the primary entry points for potentially untrusted data.
Boundary markers: The skill does not define specific delimiters or instructions for the agent to ignore embedded commands within captured context.
Capability inventory: The skill does not define or request capabilities for arbitrary command execution, network access, or direct file system manipulation beyond the specialized context tools.
Sanitization: There is no mention of input sanitization or validation for the content stored in the context engine.

contextengine