Task framing, rather than the agent framework, is the primary driver of vulnerability in indirect prompt-injection attacks on LLM agents. This study presents a controlled empirical analysis across 3,274trials, benchmarking LangChain, MCP, and a Direct Anthropic API baseline exclusively on ClaudeSonnet 4 to isolate framework-specific vulnerabilities. The results demonstrate that vulnerabilityis framework-independent on the model tested: MCP adds no measurable attack surface relative toa raw API baseline (p > 0.6). Task framing was the dominant factor, with instruction-followingtasks reaching 89-91% attack success rates versus 15-26% for constrained tasks. LangChain’sFAISS retrieval introduced task-dependent exposure, amplifying attack surface when payloads weresemantically similar to the query and inadvertently suppressing them when they were not. Hardenedsystem prompts reduced attack success by over 17× but created a brittle threshold rather than agradient: among the 20-21 valid trials where the model engaged despite hardening, exploitation ratesexceeded 55%.