The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to read external project files which serves as a potential vector for indirect prompt injection if the files contain malicious content.
Ingestion points: The skill reads from .claude/project-context.md or .cursor/project-context.md during the 'Initial Assessment' phase.
Boundary markers: No specific delimiters or instructions to ignore potential commands within these files are provided.
Capability inventory: The skill does not possess active capabilities such as network access, shell command execution, or file writing.
Sanitization: There are no defined processes for sanitizing or validating the content read from the project context files.

content-strategy