704-technologies-sql
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute build commands (
mvn compile,./mvnw compile,mvn clean verify, and./mvnw clean verify) as a mandatory step before proposing or promoting changes. This grants the agent the capability to execute code defined in the workspace's build configuration (e.g., Maven plugins or test suites). - [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and acts upon untrusted data from the workspace while possessing powerful execution capabilities.
- Ingestion points: The agent is instructed to read schema files, migrations, SQL queries, tests, and database configuration from the workspace (
SKILL.md, Workflow Step 1). - Boundary markers: There are no instructions to use delimiters or ignore embedded commands within the ingested workspace files.
- Capability inventory: The skill mandates the execution of shell-based build and test commands (
mvn,./mvnw) which can trigger arbitrary code execution via compromised build descriptors (pom.xml) or test files. - Sanitization: No validation or sanitization of project files or build configurations is performed before the agent triggers the build commands.
Audit Metadata