context-management-context-restore
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes a process for retrieving and 'rehydrating' context from external sources like file systems and vector databases via the
context_sourceparameter. This creates a potential vulnerability where untrusted data within those sources could contain malicious instructions designed to influence the agent's behavior during the reconstruction process. - Ingestion points: The skill ingests data from external
context_source(file system or vector database) to reconstruct project states. - Boundary markers: The instructions lack defined delimiters or clear instructions for the agent to ignore potential instructions embedded within the retrieved context components.
- Capability inventory: The described system has the capability to read files (
load_component) and perform semantic searches to build the active context. - Sanitization: The provided logic does not include sanitization, filtering, or validation steps for the retrieved semantic content before it is processed by the AI agent.
Audit Metadata