The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests potentially untrusted data from session debug logs and daemon logs. The analysis identifies the following: Ingestion points include {debug_log_path} and {user_home}/.claude/daemon.log. Boundary markers are absent. The capability inventory includes diagnostic reporting and subagent invocation. Sanitization of log content is not specified. While this presents an attack surface, the risk is inherent to diagnostic tools and the skill does not possess high-risk capabilities like network exfiltration.
[DATA_EXPOSURE]: The skill accesses sensitive application settings at {user_home}/.claude/settings.json and project directories. This access is consistent with the skill's primary purpose of environment diagnostics and software error troubleshooting.

debug