AAK-TRUST-003: Wildcard or overly broad permission allows
- Severity
- HIGH
- Category
- Trust Boundaries (
trust-boundary) - Scanner
agent_audit_kit/scanners/trust_boundary.py- Auto-fixable
- No.
- SARIF name
WildcardPermissionAllow
What it detects
Permission allow patterns use wildcards (*, **) or broad tool names. This bypasses the permission system, allowing unchecked tool execution.
How to fix
Use narrowest possible permission patterns. Specify exact tool names and path constraints.
To confirm the finding is gone, report only this rule:
agent-audit-kit scan . --rules AAK-TRUST-003
Limitations
Matches only the exact strings * (or any run of asterisks), mcp__*, Bash(*), Edit(**), Read(**) and Write(**) in permissions.allow. A bare tool name (Bash) or a prefix grant (Bash(sudo:*)) is not matched.
References
- OWASP MCP Top 10: MCP02:2025 (Privilege Escalation via Scope Creep)
- OWASP Agentic Top 10: ASI02 (Tool Misuse and Exploitation)
- Adversa AI MCP Top 25 (AAK-defined ids): ADV-SCOPE-04 (Wildcard Permissions)
- CSA AI Controls Matrix (AICM): IAM-02
All rules. This page is generated from the rule registry when the documentation is built; to change it, change the rule in agent_audit_kit/rules/builtin.py. To give the rule a hand-written write-up, add docs/rules/AAK-TRUST-003.md: the page then shows that text, with the registry reference appended under it.