prompt-architect
Pass
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted user input (prompts for critique or rewriting) which can contain embedded instructions designed to subvert the agent's behavior.
- Ingestion points: User-provided prompts are ingested in 'Critique', 'Rewrite', and 'Compare' modes as described in SKILL.md and references/prompt-architect-3.0.md.
- Boundary markers: The skill lacks instructions for the agent to use secure delimiters (such as specific XML tags or unique string wrappers) or to ignore instructions found within the input text.
- Capability inventory: No executable scripts, network access, or file system operations are defined within the skill itself, which limits the potential impact of an injection to the current conversation context.
- Sanitization: There is no logic provided for sanitizing, escaping, or validating the content of the prompts submitted by users.
Audit Metadata