unslop-ai-codebase
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_CONTEXT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase content (comments, docstrings, READMEs) to perform rewrites, creating a surface for instructions embedded in those files to influence the agent.
- Ingestion points: Target files' comments, docstrings, and prose documentation.
- Boundary markers: The skill lacks explicit delimiters or instructions to the agent to ignore instructions embedded in the processed data.
- Capability inventory: The agent has file read/write access and the ability to execute the provided check.sh validation script.
- Sanitization: No sanitization is performed on input text before processing.
- [DYNAMIC_CONTEXT_INJECTION]: The command definitions utilize shell commands to provide repository status to the agent at load time.
- Evidence: Commands unslop.md and unslop-diff.md execute
git diff --stat HEAD && git status --shortvia the dynamic execution syntax. - Analysis: This usage is restricted to gathering local repository metadata and is consistent with standard developer workflows, representing a functional use of the feature rather than a malicious one.
Audit Metadata