cursor-principle-encode-lessons-in-structure
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to capture human interventions and test failures to create structural enforcement mechanisms such as scripts, lint rules, or runtime checks. This creates an attack surface where malicious feedback could be persisted into automated tooling or configuration.
- Ingestion points: The skill explicitly mentions capturing human corrections and test results as signals in
SKILL.md. - Boundary markers: No explicit delimiters or instructions to ignore or sanitize embedded instructions within the feedback are provided.
- Capability inventory: The skill encourages the generation of scripts and enforcement mechanisms, which implies the use of file-writing, linting, or shell execution capabilities if available to the agent.
- Sanitization: The instructions do not include validation, escaping, or filtering of the input corrections before they are encoded into structure.
- [EXTERNAL_DOWNLOADS]: The
MIRROR.mdfile references source code and documentation from a public repository atgithub.com/cursor/plugins.git. This reference tracks the skill's provenance from the Cursor project's official plugin repository.
Audit Metadata