skill-evolver
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for ingesting external data such as competitor research and session history to inform skill modifications, creating a potential injection surface.
- Ingestion points: Synthesis of competitor mechanisms and deep research findings are incorporated into the decision-making process.
- Boundary markers: The reference documentation (evidence-to-change.md) explicitly mandates the use of 'prepared safe inputs' and 'distillation' before processing.
- Capability inventory: The skill has the ability to modify source code through the 'skill-creator' specialist tool.
- Sanitization: A specific redaction and distillation route is defined for conversation history to ensure inputs are safe for agent consumption.
- [EXTERNAL_DOWNLOADS]: The skill references a public catalog and installation guide hosted on the author's official GitHub repository (github.com/daymade/claude-code-skills). These references are used for identifying legitimate plugin packages and do not involve unauthorized code execution.
Audit Metadata