skills/oodle-ai/agent-skills/oodle/Gen Agent Trust Hub

oodle

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [SAFE]: No malicious patterns or critical security risks were identified. The skill instructions follow prescriptive safety guidelines, emphasizing user confirmation for destructive actions and strict data sanitization rules.\n- [PROMPT_INJECTION]: The oodle-onboarding skill (in skills/oodle-onboarding/SKILL.md) exhibits an indirect prompt injection surface by processing remote integration setup specifications to execute commands. \n
  • Ingestion points: Fetches setup specifications from the Oodle API via oodle integrations get-setup-spec <type>.\n
  • Boundary markers: Mitigated by instructions requiring the agent to display the command and obtain explicit user confirmation before every execution.\n
  • Capability inventory: Allows execution of shell commands such as kubectl, helm, and other platform configuration tools.\n
  • Sanitization: Prescriptive rules are included for redacting API keys and sensitive tokens before displaying commands or configurations to the user.\n- [EXTERNAL_DOWNLOADS]: The skill documentation guides the installation of the oodle CLI tool and agent skills using official package managers (Homebrew, Go, npx) and vendor-owned repositories (github.com/oodle-ai/*). These downloads are verified to originate from official vendor infrastructure.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 10:02 AM
Security Audit — agent-trust-hub — oodle