fewer-permission-prompts
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [DATA_EXFILTRATION]: The skill scans session transcript files located in
~/.claude/projects/across all projects to analyze tool usage. These files contain history that may include sensitive user data or command outputs.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from previous session transcripts, creating an attack surface where past malicious content could attempt to influence the allowlist logic.\n - Ingestion points: Session transcripts in
~/.claude/projects/(SKILL.md)\n - Boundary markers: None identified in instructions.\n
- Capability inventory: File system reading (transcripts), file system writing (.claude/settings.json).\n
- Sanitization: Explicit instructions to filter for read-only commands and strictly block arbitrary code execution (ACE) patterns.\n- [COMMAND_EXECUTION]: The skill modifies the project's security configuration in
.claude/settings.json. It contains explicit safeguards to ensure only read-only commands are allowlisted and specifically forbids patterns that would enable arbitrary code execution, such as shell interpreters or broad wildcards.
Audit Metadata