typo3-upgrade-baseline
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interacts with external websites to discover URLs, inventory components, and capture DOM/pixel data (Workflow steps 2, 4, and 6). This process ingests untrusted data from the site under test into the agent's context. While the skill focuses on technical measurement and comparison (strict-zero self-test), there are no explicit instructions for sanitizing the ingested HTML or DOM content or using boundary markers to prevent the agent from interpreting embedded instructions. An attacker who controls the site being baselined could attempt to embed malicious instructions in the page content to influence the agent's behavior during the analysis phase.
- Ingestion points: Site URL discovery (Step 2) and component/journey inventory (Step 4) in SKILL.md.
- Boundary markers: None specified for the processed site content.
- Capability inventory: Uses the
t3uCLI tool for environment checks (t3u doctor) and node management (t3u node-brief), along with automated browser captures. - Sanitization: No explicit sanitization or filtering of external site content is described.
Audit Metadata