The Agent Skills Directory

[SAFE]: The skill implements a routing table using keyword and intent matching to delegate tasks to other agents (e.g., cs-cfo-advisor, cs-ciso-advisor). No patterns of prompt injection, behavior overrides, or malicious instructions were identified.
[SAFE]: Data processing is contained within the local environment. The skill accesses local context files and logs decisions to the local filesystem (~/.claude/briefs/). No network exfiltration or sensitive data exposure patterns are present.
[SAFE]: No obfuscation techniques, hardcoded credentials, or unauthorized command execution patterns (such as sudo or chmod) were found in the skill content.
[SAFE]: Regarding indirect prompt injection surfaces, the skill ingests user input via the parameter. While it lacks explicit boundary markers, its capabilities are restricted to invoking other internal advisors and logging, which is expected behavior for its primary purpose.

founder-mode