context-engineering

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • SAFE (SAFE): The skill is strictly informational, providing strategies for handling large context windows and multi-agent workflows. No executable logic is included.
  • EXTERNAL_DOWNLOADS (SAFE): Although the metadata mentions an MCP server (@clawfu/mcp-skills), there are no instructions or scripts provided to install, download, or execute external dependencies.
  • PROMPT_INJECTION (SAFE): The content focuses on 'Context Engineering' techniques such as summarization and pruning. It does not contain instructions to bypass safety guardrails or extract system prompts.
  • DATA_EXFILTRATION (SAFE): There are no file system access or network communication patterns detected in the markdown text.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:39 PM