The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions direct the agent to access the ~/.claude/projects/ directory to log telemetry data in a file named skill-telemetry.md. This directory is a sensitive path that stores internal project history and state. While the operation is described as a local log append, interactions with sensitive configuration directories carry inherent risks of data exposure if the agent's context is compromised.
[PROMPT_INJECTION]: The skill exhibits an Indirect Prompt Injection surface because it is designed to ingest and analyze untrusted SKILL.md files from a registry.
Ingestion points: The agent scans and reads the contents of all SKILL.md files within the provided registry root or category subdirectories.
Boundary markers: There are no explicit instructions for the agent to use delimiters or to treat the content of the scanned files as data that should not be interpreted as instructions.
Capability inventory: The skill possesses file reading capabilities for the scan and file writing capabilities to auto-fix frontmatter and generate README files.
Sanitization: No sanitization or filtering logic is specified for the content read from the external skill files, which could allow a malicious skill file to influence the agent's reasoning during the validation phases.

registry-sync