• There are 183 files with 51,430 lines of code.
• 9 very long files (14,836 lines of code)
• 25 long files (17,639 lines of code)
• 41 medium size files (12,677 lines of code)
• 19 small files (2,800 lines of code)
• 89 very small files (3,478 lines of code)

File	# lines	≈ tokens to read	# changes (1y)	≈ lines edited per change	lines read per line changed
registry.yaml in root	977	~5k–15k	71	~30	~30
guide.ipynb in capabilities/contextual-embeddings	1851	~10k–30k	13	~500	~3
session_memory_compaction.ipynb in misc	1474	~10k–30k	14	~700	~2
automatic-context-compaction.ipynb in tool_use	1248	~7k–20k	13	~500	~3
07_Hosting_the_agent.ipynb in claude_agent_sdk	2160	~15k–50k	4	~1,000	~2
prompt_caching.ipynb in misc	640	~3k–10k	12	~150	~5
tool_search_with_embeddings.ipynb in tool_use	1090	~7k–20k	7	~700	~2
prompting_for_frontend_aesthetics.ipynb in coding	1175	~7k–20k	6	~500	~2
CMA_with_mongodb_atlas.ipynb in managed_agents	1663	~10k–30k	4	~700	~2
CMA_remember_user_preferences.ipynb in managed_agents	649	~3k–10k	10	~150	~5
sre_incident_responder.ipynb in managed_agents	850	~5k–15k	7	~200	~5
authors.yaml in root	160	~1k–3k	35	~10	~15
using_citations.ipynb in misc	887	~5k–15k	6	~200	~5
validate_all_notebooks.py in scripts	632	~3k–10k	8	~200	~3
generate_test_cases.ipynb in misc	792	~5k–15k	6	~200	~5
tool_choice.ipynb in tool_use	745	~5k–15k	6	~200	~3
CMA_plan_big_execute_small.ipynb in managed_agents	536	~3k–10k	8	~300	~2
CMA_coordinate_specialist_team.ipynb in managed_agents	532	~3k–10k	7	~300	~2
orchestrator_workers.ipynb in patterns/agents	585	~3k–10k	6	~300	~2
extended_thinking_with_tool_use.ipynb in extended_thinking	862	~5k–15k	4	~200	~5

Tokens are lines × 7 to 14 (analysis.fileReadsForChanges). Lines edited per change are the file's lines added and deleted divided by its commits over the whole history. Agents read about 2,000 lines at a time: when a 17,000-line file was split into modules, the tokens an agent read for the same change fell by 83% (The Economic Benefit of Refactoring, 2026).

There are 71 files with lines longer than 120 characters. In total, there are 1720 long lines.

File	# lines	# units	# long lines
threat_intel_enrichment_agent.ipynb in tool_use	1895	-	144
session_memory_compaction.ipynb in misc	1474	-	103
prompting_for_frontend_aesthetics.ipynb in coding	1175	-	90
08_Dynamic_workflows.ipynb in claude_agent_sdk	949	-	86
07_Hosting_the_agent.ipynb in claude_agent_sdk	2160	-	76
using_citations.ipynb in misc	887	-	68
automatic-context-compaction.ipynb in tool_use	1248	-	66
reproduce_agentic_search_benchmarks.ipynb in evals/agentic_search	788	-	63
guide.ipynb in capabilities/contextual-embeddings	1851	-	55
04_migrating_from_openai_agents_sdk.ipynb in claude_agent_sdk	885	-	53
guide.ipynb in fable_5_fallback_billing	563	-	46
building_evals.ipynb in misc	407	-	44
admin_api.ipynb in misc	800	-	38
tool_choice.ipynb in tool_use	745	-	35
CMA_remember_user_preferences.ipynb in managed_agents	649	-	35
CMA_consult_an_advisor.ipynb in managed_agents	592	-	34
05_Building_a_session_browser.ipynb in claude_agent_sdk	542	-	33
tool_search_with_embeddings.ipynb in tool_use	1090	-	32
CMA_plan_big_execute_small.ipynb in managed_agents	536	-	31
customer_service_api.py in tool_use/utils	289	6	30
orchestrator_workers.ipynb in patterns/agents	585	-	30
sre_incident_responder.ipynb in managed_agents	850	-	30
CMA_operate_in_production.ipynb in managed_agents	403	-	30
reading_charts_graphs_powerpoints.ipynb in multimodal	508	-	30
CMA_coordinate_specialist_team.ipynb in managed_agents	532	-	26
generate_test_cases.ipynb in misc	792	-	24
batch_processing.ipynb in misc	462	-	23
CMA_use_skills_from_a_repo.ipynb in managed_agents	351	-	21
data_analyst_agent.ipynb in managed_agents	413	-	21
jsonl support_tickets.jsonl in managed_agents/example_data/prompt_versioning_and_rollback	20	-	20
customer_service_agent.ipynb in tool_use	404	-	18
prompt_caching.ipynb in misc	640	-	18
CMA_prompt_versioning_and_rollback.ipynb in managed_agents	447	-	18
building_moderation_filter.ipynb in misc	366	-	15
async_multi_agent_orchestration.ipynb in patterns/agents	564	-	15
slack_data_bot.ipynb in managed_agents	369	-	15
tool_search_alternate_approaches.ipynb in tool_use	452	-	14
CMA_gate_human_in_the_loop.ipynb in managed_agents	312	-	13
CMA_orchestrate_issue_to_pr.ipynb in managed_agents	266	-	13
CMA_with_mongodb_atlas.ipynb in managed_agents	1663	-	12
extended_thinking_with_tool_use.ipynb in extended_thinking	862	-	12
CMA_pin_inference_geo.ipynb in managed_agents	258	-	11
jsonl samples.jsonl in capabilities/content_moderation/data/ad_creatives	10	-	10
parallel_tools.ipynb in tool_use	352	-	9
pdf_upload_summarization.ipynb in misc	260	-	9
CMA_explore_unfamiliar_codebase.ipynb in managed_agents	278	-	8
research_corpus.py in tool_use/context_engineering	275	1	7
read_web_pages_with_haiku.ipynb in misc	135	-	7
basic_workflows.ipynb in patterns/agents	797	-	7
evaluation.xml in tool_evaluation	34	-	7

• File size measurements show the distribution of size of files.
• Files are classified in four categories based on their size (lines of code): 1-100 (very small files), 101-200 (small files), 201-500 (medium size files), 501-1000 (long files), 1001+(very long files).
• It is a good practice to keep files small. Long files may become "bloaters", code that have increased to such gargantuan proportions that they are hard to work with.