deal-desk-approval
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security findings were identified. The skill is a collection of Markdown instructions and reference files for designing deal desk approval chains.
- [PROMPT_INJECTION]: The skill uses instructional steering to ensure the AI follows specific domain guardrails, such as refusing to present unverified industry numbers as facts and avoiding legal or accounting advice. These are safety-oriented instructions and do not attempt to bypass agent security filters or extract system prompts.
- [DATA_EXFILTRATION]: No network operations, credential harvesting, or sensitive file path access patterns were detected. The skill does not request access to external tools or environments.
- [DYNAMIC_CONTEXT_INJECTION]: The skill was scanned for the dynamic context injection pattern (
!command). No instances of load-time shell execution or command injection were found. - [INDIRECT_PROMPT_INJECTION]: While the skill defines an interview process to ingest user deal data (segment mix, ACV, discount distribution), it lacks high-privilege capabilities (file system writes, network requests, or code execution) that could be exploited via untrusted data. The skill emphasizes data provenance and strict boundary markers for what constitutes a valid 'Deal Desk Approval Spec'.
Audit Metadata