teleport-acl-lifecycle

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill includes explicit defensive instructions to ignore and flag any instruction-like text found in resource metadata (e.g., "ignore previous instructions"). The static detector's hit is a false positive based on these security guidelines.
  • [COMMAND_EXECUTION]: The skill implements a strict 'Read Capture Discipline' and a command allowlist. It uses robust shell-escaping for all interpolated variables and mandates the use of argument-passing for parsers like jq and python3, effectively preventing command injection.
  • [DATA_EXFILTRATION]: No network exfiltration or sensitive data exposure patterns were detected. The skill relies on existing local credentials and manages all data within private temporary directories that are cleaned up after use.
  • [INDIRECT_PROMPT_INJECTION]: The skill acknowledges the risk of processing untrusted data from Teleport resources. It mitigates this via mandatory evidence chains: it identifies ingestion points (CLI output), mandates human approval for all writes, and provides specific instructions for the agent to recognize and ignore malicious instructions embedded in data.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 04:49 PM
Security Audit — agent-trust-hub — teleport-acl-lifecycle