The Agent Skills Directory

[PROMPT_INJECTION]: The skill identifies a potential indirect prompt injection surface through its state management implementation.
Ingestion points: The skill instructions in SKILL.md require the agent to read data from .claude/context/memory/learnings.md via the cat command at the start of every session.
Boundary markers: There are no delimiters, XML tags, or specific instructions to the agent to ignore potentially malicious embedded commands within the ingested file content.
Capability inventory: The skill grants the agent access to high-impact tools including Bash, Write, Edit, Read, Grep, and Glob across all scripts.
Sanitization: The skill lacks any logic to validate, escape, or sanitize the content of the memory file before it is processed by the agent.
[COMMAND_EXECUTION]: The skill configuration and instructions explicitly require the use of system commands.
Evidence: The Memory Protocol section in SKILL.md contains a mandatory instruction to execute cat .claude/context/memory/learnings.md.
Tools: The skill frontmatter in SKILL.md explicitly enables the Bash tool, allowing the agent to perform arbitrary shell operations.

solidjs-expert