context-window-management
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists solely of markdown instructions and a configuration schema. No code, scripts, or external dependencies are included. The state management defined in STATE_SCHEMA.yaml uses standard fields for tracking session compaction.
- [INDIRECT_PROMPT_INJECTION]: The instructions involve processing and summarizing conversation history, which creates a surface for indirect prompt injection. This risk is inherent to context management tasks and is not escalated by the skill's logic. * Ingestion points: Session context, code blocks, and task details described in SKILL.md. * Boundary markers: None present. * Capability inventory: Local file system write operations for state tracking and handoff documents. * Sanitization: None present.
Audit Metadata