New session starts 32% full due to preloaded system/MCP/memory payload

Author: bcihancCreated Aug 24, 2025Updated Mar 28, 2026

In a fresh SuperClaude session with claude-sonnet-4-20250514 (200k), the context is already 32% used (~64k/200k) before I type anything—coming from preloaded system prompt (~2.8k), system tools (~12.3k), MCP tools (~37.6k), memory files (~10.5k), and custom agents (~374)—which unnecessarily reduces available headroom, increases token cost/latency, and makes larger tasks hit limits much sooner; this default preload feels excessive and frustrating.

bash
> /context 
  ⎿  ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁   Context Usage
     ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁   claude-sonnet-4-20250514 • 64k/200k tokens (32%)
     ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛀ ⛁ ⛁ ⛁ 
     ⛁ ⛁ ⛀ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ System prompt: 2.8k tokens (1.4%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ System tools: 12.3k tokens (6.2%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ MCP tools: 37.6k tokens (18.8%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ Custom agents: 374 tokens (0.2%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ Memory files: 10.5k tokens (5.3%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ Messages: 8 tokens (0.0%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛶ Free space: 136.4k (68.2%)

     MCP tools · /mcp
     └ mcp__sequential-thinking__sequentialthinking (sequential-thinking)1.3k tokens

     └ mcp__context7__resolve-library-id (context7): 691 tokens
     └ mcp__context7__get-library-docs (context7): 652 tokens
     └ mcp__playwright__browser_close (playwright): 390 tokens
     └ mcp__playwright__browser_resize (playwright): 439 tokens
     └ mcp__playwright__browser_console_messages (playwright): 393 tokens
     └ mcp__playwright__browser_handle_dialog (playwright): 443 tokens
     └ mcp__playwright__browser_evaluate (playwright): 488 tokens
     └ mcp__playwright__browser_file_upload (playwright): 440 tokens
     └ mcp__playwright__browser_install (playwright): 408 tokens
     └ mcp__playwright__browser_press_key (playwright): 437 tokens
     └ mcp__playwright__browser_type (playwright): 545 tokens
     └ mcp__playwright__browser_navigate (playwright): 415 tokens
     └ mcp__playwright__browser_navigate_back (playwright): 395 tokens
     └ mcp__playwright__browser_navigate_forward (playwright): 395 tokens
     └ mcp__playwright__browser_network_requests (playwright): 397 tokens
     └ mcp__playwright__browser_take_screenshot (playwright): 631 tokens
     └ mcp__playwright__browser_snapshot (playwright): 401 tokens
     └ mcp__playwright__browser_click (playwright): 516 tokens
     └ mcp__playwright__browser_drag (playwright): 526 tokens
     └ mcp__playwright__browser_hover (playwright): 453 tokens
     └ mcp__playwright__browser_select_option (playwright): 501 tokens
     └ mcp__playwright__browser_tab_list (playwright): 392 tokens
     └ mcp__playwright__browser_tab_new (playwright): 426 tokens
     └ mcp__playwright__browser_tab_select (playwright): 420 tokens
     └ mcp__playwright__browser_tab_close (playwright): 420 tokens
     └ mcp__playwright__browser_wait_for (playwright): 467 tokens
     └ mcp__magic__21st_magic_component_builder (magic): 744 tokens
     └ mcp__magic__logo_search (magic): 709 tokens
     └ mcp__magic__21st_magic_component_inspiration (magic): 556 tokens
     └ mcp__magic__21st_magic_component_refiner (magic): 665 tokens
     └ mcp__serena__read_file (serena): 651 tokens
     └ mcp__serena__create_text_file (serena): 467 tokens
     └ mcp__serena__list_dir (serena): 575 tokens
     └ mcp__serena__find_file (serena): 496 tokens
     └ mcp__serena__replace_regex (serena): 802 tokens
     └ mcp__serena__search_for_pattern (serena): 1.4k tokens
     └ mcp__serena__get_symbols_overview (serena): 564 tokens
     └ mcp__serena__find_symbol (serena): 1.6k tokens
     └ mcp__serena__find_referencing_symbols (serena): 676 tokens
     └ mcp__serena__replace_symbol_body (serena): 539 tokens
     └ mcp__serena__insert_after_symbol (serena): 555 tokens
     └ mcp__serena__insert_before_symbol (serena): 562 tokens
     └ mcp__serena__write_memory (serena): 470 tokens
     └ mcp__serena__read_memory (serena): 488 tokens
     └ mcp__serena__list_memories (serena): 387 tokens
     └ mcp__serena__delete_memory (serena): 442 tokens
     └ mcp__serena__execute_shell_command (serena): 637 tokens
     └ mcp__serena__activate_project (serena): 419 tokens
     └ mcp__serena__switch_modes (serena): 442 tokens
     └ mcp__serena__check_onboarding_performed (serena): 417 tokens
     └ mcp__serena__onboarding (serena): 406 tokens
     └ mcp__serena__think_about_collected_information (serena): 436 tokens
     └ mcp__serena__think_about_task_adherence (serena): 436 tokens
     └ mcp__serena__think_about_whether_you_are_done (serena): 401 tokens
     └ mcp__serena__prepare_for_new_conversation (serena): 392 tokens
     └ mcp__morphllm-fast-apply__read_file (morphllm-fast-apply): 547 tokens
     └ mcp__morphllm-fast-apply__read_multiple_files (morphllm-fast-apply)477 tokens

     └ mcp__morphllm-fast-apply__write_file (morphllm-fast-apply): 462 tokens
     └ mcp__morphllm-fast-apply__tiny_edit_file (morphllm-fast-apply): 591 tokens
     └ mcp__morphllm-fast-apply__create_directory (morphllm-fast-apply): 458 tokens
     └ mcp__morphllm-fast-apply__list_directory (morphllm-fast-apply): 460 tokens
     └ mcp__morphllm-fast-apply__list_directory_with_sizes                504 tokens
     (morphllm-fast-apply):
     └ mcp__morphllm-fast-apply__directory_tree (morphllm-fast-apply): 484 tokens
     └ mcp__morphllm-fast-apply__move_file (morphllm-fast-apply): 476 tokens
     └ mcp__morphllm-fast-apply__search_files (morphllm-fast-apply): 510 tokens
     └ mcp__morphllm-fast-apply__get_file_info (morphllm-fast-apply): 456 tokens
     └ mcp__morphllm-fast-apply__list_allowed_directories                 400 tokens
     (morphllm-fast-apply):
     └ mcp__morphllm-fast-apply__edit_file (morphllm-fast-apply): 985 tokens

     Custom agents · /agents
     └ python-expert (Project): 31 tokens
     └ system-architect (Project): 27 tokens
     └ refactoring-expert (Project): 29 tokens
     └ devops-architect (Project): 25 tokens
     └ learning-guide (Project): 27 tokens
     └ security-engineer (Project): 24 tokens
     └ frontend-architect (Project): 25 tokens
     └ quality-engineer (Project): 23 tokens
     └ root-cause-analyst (Project): 30 tokens
     └ socratic-mentor (Project): 31 tokens
     └ performance-engineer (Project): 24 tokens
     └ requirements-analyst (Project): 25 tokens
     └ backend-architect (Project): 25 tokens
     └ technical-writer (Project): 28 tokens

     Memory files · /memory
     └ User (/Users/cihan/.claude/CLAUDE.md): 234 tokens
     └ User (/Users/cihan/.claude/FLAGS.md): 1.2k tokens
     └ User (/Users/cihan/.claude/PRINCIPLES.md): 575 tokens
     └ User (/Users/cihan/.claude/RULES.md): 3.4k tokens
     └ User (/Users/cihan/.claude/MODE_Brainstorming.md): 456 tokens
     └ User (/Users/cihan/.claude/MODE_Introspection.md): 419 tokens
     └ User (/Users/cihan/.claude/MODE_Orchestration.md): 415 tokens
     └ User (/Users/cihan/.claude/MODE_Task_Management.md): 957 tokens
     └ User (/Users/cihan/.claude/MODE_Token_Efficiency.md): 884 tokens
     └ User (/Users/cihan/.claude/MCP_Context7.md): 303 tokens
     └ User (/Users/cihan/.claude/MCP_Magic.md): 303 tokens
     └ User (/Users/cihan/.claude/MCP_Morphllm.md): 331 tokens
     └ User (/Users/cihan/.claude/MCP_Playwright.md): 309 tokens
     └ User (/Users/cihan/.claude/MCP_Sequential.md): 330 tokens
     └ User (/Users/cihan/.claude/MCP_Serena.md): 362 tokens

Source: SuperClaude-Org/SuperClaude_Framework