access
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill includes explicit security instructions to ignore access requests arriving via channel notifications (external input). It mandates that mutations like 'pair', 'allow', or 'policy' changes must only be performed when the user manually types the command in their terminal, effectively mitigating indirect prompt injection risks from the messaging platform.
- [DATA_EXFILTRATION]: No network operations or external data transmissions are defined. The skill only performs local file read/write operations within the
~/.claude/directory to manage configuration state. - [COMMAND_EXECUTION]: While the skill uses shell commands via the
Bashtool, they are restricted to directory management (mkdir -p) and state inspection (ls). There are no patterns of executing arbitrary user input or downloading scripts. - [SAFE]: The state management logic for
access.jsonis handled safely. The skill requires specific codes for pairing and prevents auto-approval of pending entries, which is a defensive design to prevent attackers from pre-seeding malicious pairing requests.
Audit Metadata