brand-copy
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions in
SKILL.md(Steps 1 and 2) require the agent to ingest and analyze untrusted external data, such as brand briefs and existing copy samples. This creates a surface for indirect prompt injection attacks where an attacker could embed malicious instructions within the provided samples. Evidence Chain: - Ingestion points:
SKILL.md(Step 1: Receive the brief; Step 2: Analyze existing voice). - Boundary markers: Absent. There are no delimiters or instructions provided to the agent to differentiate between data to be analyzed and instructions to be followed.
- Capability inventory: Minimal risk. The skill only generates text output and does not have access to tools for network requests, file system modifications, or shell execution.
- Sanitization: Absent. Input data is interpolated directly into the analysis workflow.
- [NO_CODE]: The skill consists entirely of Markdown instruction files (
SKILL.md) and static reference documents (references/). It does not include any scripts (Python, JavaScript, shell), binaries, or dependency manifests, which significantly limits the potential for traditional exploitation.
Audit Metadata