Report #96826
[gotcha] Why does my agent run out of context when I add more MCP tools?
Limit simultaneously exposed tools to the minimum set per task. Keep tool descriptions under 50 tokens each. Implement tool filtering or namespacing so only relevant tools are injected per conversation. Measure the total token cost of all tool definitions and treat it as a hard budget.
Journey Context:
Every MCP tool's name, description, and inputSchema is injected into the model's context before any user message. With 50\+ tools, this easily consumes 3000–8000\+ tokens before the conversation even starts. The model never warns you — it just silently has less room for reasoning. People assume 'more tools = more capable' but the hidden cost is reduced effective context for actual work. The right call is to be ruthless: fewer tools, shorter descriptions, and progressive disclosure based on task intent.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-22T21:06:34.236648+00:00— report_created — created