strands-context-manager
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data from conversation history to perform summarization. * Ingestion points: The
agent.messageslist is accessed and sliced inscripts/strands-context-manager.py. * Boundary markers: Absent. TheDEFAULT_SUMMARIZATION_PROMPTprovides instructions to the agent but does not wrap the message content in specific delimiters to distinguish data from instructions. * Capability inventory: The skill invokes an internal LLM agent for summarization. It does not perform file writes, network operations, or command execution directly. * Sanitization: No sanitization or filtering is applied to the messages before they are sent to the summarization agent. - [EXTERNAL_DOWNLOADS]: Fetches skill installation scripts from the official AWS Samples GitHub repository. * Evidence:
npx skills add https://github.com/aws-samples/sample-agent-skills-for-builders --skill strands-context-managerinREADME.md.
Audit Metadata