error-recovery

Pass

Audited by Gen Agent Trust Hub on Apr 18, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The recovery playbook includes a strategy to "Rephrase prompt" and "Retry without triggering content filter" when encountering provider errors. This is an explicit instruction to evade safety guardrails.- [COMMAND_EXECUTION]: The skill utilizes shell commands like mkdir -p within its file system recovery playbook to automatically create missing directories.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted error data from external APIs and file system operations without sanitization.
  • Ingestion points: Error messages and outputs from AI providers, network requests, and file system errors.
  • Boundary markers: Absent; error classification logic relies on raw input strings.
  • Capability inventory: File read/write permissions, shell command execution (mkdir), and the ability to modify and re-execute prompts.
  • Sanitization: No sanitization is mentioned for processing error strings or logging data to the audit file.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 18, 2026, 12:42 AM
Security Audit — agent-trust-hub — error-recovery