consistent-lexicon

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill's primary function is to help organize and document domain-specific language. It restricts its operations to scanning the current conversation and updating a single local file (LEXICON.md). There are no commands related to network exfiltration, credential access, or privilege escalation.
  • [PROMPT_INJECTION]: The skill processes untrusted conversation data to generate its output, creating a surface for indirect prompt injection. 1. Ingestion points: The skill reads from the current conversation history and the existing LEXICON.md file. 2. Boundary markers: None are defined to differentiate between conversation content and administrative instructions. 3. Capability inventory: The skill uses file read and write operations via the agent's available tools. 4. Sanitization: The skill does not implement sanitization or validation of the conversation text before processing it into the lexicon.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 07:58 PM