find-duplicate-cases
Pass
Audited by Gen Agent Trust Hub on Mar 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands such as
findto discover test-related files (Markdown, Feature, and JS/TS specs) within the repository. - [PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection because it ingests and processes the content of various test files while possessing file modification and deletion capabilities.
- Ingestion points: Manual and automated test files (e.g.,
*.md,*.feature,*.spec.ts) are read and parsed to extract test metadata and steps (SKILL.md, Workflow Step 1). - Boundary markers: No explicit delimiters or boundary instructions are used to distinguish test data from agent instructions.
- Capability inventory: The skill is capable of performing
Merge(rewriting files) andRemove(deleting files) actions, which impacts repository integrity (SKILL.md, Workflow Step 4). - Sanitization: The skill lacks sanitization or validation of the ingested test content before it is processed or used in file-writing operations.
Audit Metadata