velt-rewriter-best-practices

Pass

Audited by Gen Agent Trust Hub on May 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains instructions in AGENTS.md to override default model behavior by prioritizing retrieval-led reasoning over pre-trained knowledge.
  • [PROMPT_INJECTION]: The skill documents a surface for indirect prompt injection by instructing the agent to process untrusted user text selections through AI generation tools.
  • Ingestion points: User-selected text captured via the textSelected event (documented in rules/shared/api/api-on-text-selected.md).
  • Boundary markers: No explicit delimiter or 'ignore' instructions are provided in the prompt interpolation examples.
  • Capability inventory: The skill utilizes replaceText for DOM modification and addComment for persistent annotation creation (documented in rules/shared/api/api-replace-text.md and rules/shared/api/api-add-comment.md).
  • Sanitization: No sanitization or escaping of external content is implemented in the documented code patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
May 16, 2026, 09:03 PM
Security Audit — agent-trust-hub — velt-rewriter-best-practices