pr
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill includes metadata that references a source file within the Humanlayer GitHub organization for attribution purposes.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and process external repository data, such as a
GLOSSARY.mdfile and code diffs, which are potential vectors for indirect prompt injection if those files contain malicious instructions. - Ingestion points: Reads local files like
GLOSSARY.md, other skill files (plain-english,deslop-written), and the project's code diff. - Boundary markers: The instructions do not specify explicit delimiters to separate user data from instructions.
- Capability inventory: The skill itself contains no commands for file modification, network communication, or code execution.
- Sanitization: No specific sanitization or filtering logic is defined for the ingested text.
Audit Metadata