audit-java
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted source code and configuration files from a repository to detect security risks.
- Ingestion points: Reads repository manifests, source files, tests, and runtime documentation (SKILL.md, Workflow Step 1).
- Boundary markers: Utilizes typed audit-findings and Observation URNs to scope output and minimize context bleed (SKILL.md, Overview).
- Capability inventory: Executes build tool wrappers and check-only repository commands using ephemeral containers (SKILL.md, Workflow Step 7).
- Sanitization: Explicitly requires redacting sensitive values and verifying reachability before claiming impact (SKILL.md, Hard Rules).
- [COMMAND_EXECUTION]: Executes repository-specific build wrappers (Maven/Gradle) and check-only validation commands.
- Evidence: Resolves project toolchains via repository-owned wrappers to determine effective build state (references/java-engineering.md).
- Mitigation: The workflow mandates the use of test-owned ephemeral containers and prohibits writing to shared or production state (SKILL.md, Workflow Step 7).
- [EXTERNAL_DOWNLOADS]: References external documentation and open-source snapshots to provide architectural guidance.
- Evidence: Includes references to official Spring documentation, OWASP security guides, and well-known open-source projects like Spring Petclinic (references/java-engineering.md).
- Context: These resources are treated as reference materials for pattern matching and architectural questions rather than code for immediate execution.
Audit Metadata