privacy-policy-malik-taiar
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection through its document analysis features.
- Ingestion points: The drafting workflow in
SKILL.md(Step 2) andSKILL.fr.md(Étape 2) instructs the agent to ingest and analyze untrusted external data sources, specifically client-provided documents such as 'T&Cs, sales conditions, order forms, and contracts,' as well as 'site/application URLs.' - Boundary markers: The skill does not implement explicit delimiters or instruction isolation techniques (like XML tagging or ignore-instructions warnings) when processing these external data sources.
- Capability inventory: The skill is primarily instructional and focused on text generation. It does not contain commands for subprocess execution, network exfiltration, or sensitive file modifications.
- Sanitization: No sanitization or filtering logic is specified for the content extracted from external documents.
- Mitigation: The risk is moderated by the skill's strict enforcement of a template-based drafting approach (Step 1 and Step 3), which requires the agent to follow a pre-validated structure and wording rather than allowing free-form generation based solely on the external input.
Audit Metadata