skills/rmyndharis/antigravity-skills/full-stack-orchestration-full-stack-feature/Gen Agent Trust Hub
full-stack-orchestration-full-stack-feature
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data via the
$ARGUMENTSplaceholder and interpolates it directly into prompts for high-capability sub-agents without sufficient isolation or sanitization. - Ingestion points: User input enters the workflow through the
$ARGUMENTSvariable, which is used in every phase of the instructions inSKILL.md(Steps 1 through 12). - Boundary markers: The skill lacks explicit delimiters or boundary markers (such as triple backticks or XML tags) around the interpolated
$ARGUMENTScontent, which could allow malicious instructions within the input to be interpreted as authoritative commands by the sub-agents. - Capability inventory: The sub-agents triggered by this skill (e.g.,
deployment-engineer,database-architect,sql-pro) have broad capabilities, including generating infrastructure-as-code, executing database migrations, and configuring CI/CD pipelines. - Sanitization: There is no evidence of input validation or sanitization applied to the
$ARGUMENTSstring before it is passed to the specialized agents.
Audit Metadata