Skills
Navigate
skills/reason-machines/ai-agent-skills/agent-skills-context-engineering/Gen Agent Trust Hub

agent-skills-context-engineering

Pass

Audited by Gen Agent Trust Hub on Oct 8, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions utilize curl to fetch additional markdown files from a third-party GitHub repository (muratcankoylan/Agent-Skills-for-Context-Engineering).\n
  • Evidence: curl -o .claude/skills/context-fundamentals.md https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/main/skills/context-fundamentals/SKILL.md\n- [INDIRECT_PROMPT_INJECTION]: The skill defines architectures that ingest untrusted user tasks and message histories without demonstrating proper safety boundaries or sanitization.\n
  • Ingestion points: The task parameter in OrchestratorAgent.process_task and the messages list in compress_with_summary in SKILL.md.\n
  • Boundary markers: Absent in provided code snippets; data is interpolated directly into prompts without delimiters or instructions to ignore embedded commands.\n
  • Capability inventory: The skill code includes file-system writing (open(path, 'w')), directory creation (os.makedirs), and network requests (Anthropic API, requests library).\n
  • Sanitization: Absent; external data is processed and used in LLM prompts and file operations without validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 8, 2026, 11:38 PM