agentic-ai-prompt-research

Name: agentic-ai-prompt-research
Author: Leonxlnx

2.5kstars1.1kforks

How to install agentic-ai-prompt-research?

leonxlnx/agentic-ai-prompt-research

$git clone https://github.com/Leonxlnx/agentic-ai-prompt-research

Installs into the current project.

Summary

A research project exploring how modern agentic AI coding assistants work under the hood. This repository contains our best understanding of the prompt architecture, agent coordination patterns, and security mechanisms that power tools like Claude Code.

Files · 1

View on GitHub

README.md

1	# Claude Code System Prompts
2
3	A research project exploring how modern agentic AI coding assistants work under the hood. This repository contains our best understanding of the prompt architecture, agent coordination patterns, and security mechanisms that power tools like Claude Code.
4
5	Everything here is based on behavioral observation, output analysis, community discussions, and publicly shared information. These are reconstructed approximations, not verbatim copies. The actual implementation may differ significantly.
6
7	## What This Project Is
8
9	This is an educational deep-dive into the design patterns behind agentic coding assistants. We analyze how these systems:
10
11	- Assemble dynamic system prompts at runtime
12	- Coordinate multiple specialized sub-agents
13	- Classify and auto-approve tool calls safely
14	- Manage context windows through intelligent compaction
15	- Handle memory, skills, and user preferences
16
17	The goal is to help AI engineers, researchers, and builders learn from these architectural patterns and apply them in their own projects.
18
19	## What This Project Is Not
20
21	This is not a leak, dump, or direct copy of any proprietary system. The prompts documented here are our best reconstructions based on observable behavior. They represent one interpretation of how these systems likely work.
22
23	## Documented Patterns
24
25	### Core Identity
26
27	\| # \| Pattern \| Description \|
28	\|---\|---------\|-------------\|
29	\| 01 \| [Main System Prompt](prompts/01_main_system_prompt.md) \| How the master prompt is dynamically assembled from modular sections \|
30	\| 02 \| [Simple Mode](prompts/02_simple_mode.md) \| Minimal prompt variant for lightweight operation \|
31	\| 03 \| [Default Agent Prompt](prompts/03_default_agent_prompt.md) \| Base instructions inherited by all sub-agents \|
32	\| 04 \| [Cyber Risk Instruction](prompts/04_cyber_risk_instruction.md) \| Security boundaries between authorized and prohibited actions \|
33
34	### Orchestration
35
36	\| # \| Pattern \| Description \|
37	\|---\|---------\|-------------\|
38	\| 05 \| [Coordinator System Prompt](prompts/05_coordinator_system_prompt.md) \| Multi-worker orchestration with phased workflows \|
39	\| 06 \| [Teammate Prompt Addendum](prompts/06_teammate_prompt_addendum.md) \| Communication protocols for multi-agent collaboration \|
40
41	### Specialized Agents
42
43	\| # \| Pattern \| Description \|
44	\|---\|---------\|-------------\|
45	\| 07 \| [Verification Agent](prompts/07_verification_agent.md) \| Adversarial testing agent that validates implementations \|
46	\| 08 \| [Explore Agent](prompts/08_explore_agent.md) \| Read-only codebase exploration with no-modify constraints \|
47	\| 09 \| [Agent Creation Architect](prompts/09_agent_creation_architect.md) \| Generates new agent configurations from requirements \|
48	\| 10 \| [Status Line Setup Agent](prompts/10_statusline_setup_agent.md) \| Terminal status line configuration across shells \|
49
50	### Security and Permissions
51
52	\| # \| Pattern \| Description \|
53	\|---\|---------\|-------------\|
54	\| 11 \| [Permission Explainer](prompts/11_permission_explainer.md) \| Risk assessment before tool approval \|
55	\| 12 \| [Auto Mode Classifier](prompts/12_yolo_auto_mode_classifier.md) \| Multi-stage security classifier for autonomous tool execution \|
56
57	### Tool Descriptions
58
59	\| # \| Pattern \| Description \|
60	\|---\|---------\|-------------\|
61	\| 13 \| [Tool-Specific Prompts](prompts/13_tool_prompts.md) \| How individual tools (Bash, Edit, Agent, etc.) describe themselves \|
62
63	### Utility Patterns
64
65	\| # \| Pattern \| Description \|
66	\|---\|---------\|-------------\|
67	\| 14 \| [Tool Use Summary](prompts/14_tool_use_summary.md) \| Generating concise labels for completed tool batches \|
68	\| 15 \| [Session Search](prompts/15_session_search.md) \| Semantic search across past conversation sessions \|
69	\| 16 \| [Memory Selection](prompts/16_memory_selection.md) \| Selecting relevant memory files for query context \|
70	\| 17 \| [Auto Mode Critique](prompts/17_auto_mode_critique.md) \| Reviewing user-written classifier rules \|
71	\| 20 \| [Session Title](prompts/20_session_title.md) \| Lightweight title generation for session management \|
72	\| 2

People who liked this also viewed

6 picks

Type

agentic-ai-prompt-research

2.5kstars1.1kforks

View on GitHub

How to install agentic-ai-prompt-research?

leonxlnx/agentic-ai-prompt-research

$git clone https://github.com/Leonxlnx/agentic-ai-prompt-research

Installs into the current project.

Summary

Files · 1

View on GitHub

README.md

1	# Claude Code System Prompts
2
3	A research project exploring how modern agentic AI coding assistants work under the hood. This repository contains our best understanding of the prompt architecture, agent coordination patterns, and security mechanisms that power tools like Claude Code.
4
5	Everything here is based on behavioral observation, output analysis, community discussions, and publicly shared information. These are reconstructed approximations, not verbatim copies. The actual implementation may differ significantly.
6
7	## What This Project Is
8
9	This is an educational deep-dive into the design patterns behind agentic coding assistants. We analyze how these systems:
10
11	- Assemble dynamic system prompts at runtime
12	- Coordinate multiple specialized sub-agents
13	- Classify and auto-approve tool calls safely
14	- Manage context windows through intelligent compaction
15	- Handle memory, skills, and user preferences
16
17	The goal is to help AI engineers, researchers, and builders learn from these architectural patterns and apply them in their own projects.
18
19	## What This Project Is Not
20
21	This is not a leak, dump, or direct copy of any proprietary system. The prompts documented here are our best reconstructions based on observable behavior. They represent one interpretation of how these systems likely work.
22
23	## Documented Patterns
24
25	### Core Identity
26
27	\| # \| Pattern \| Description \|
28	\|---\|---------\|-------------\|
29	\| 01 \| [Main System Prompt](prompts/01_main_system_prompt.md) \| How the master prompt is dynamically assembled from modular sections \|
30	\| 02 \| [Simple Mode](prompts/02_simple_mode.md) \| Minimal prompt variant for lightweight operation \|
31	\| 03 \| [Default Agent Prompt](prompts/03_default_agent_prompt.md) \| Base instructions inherited by all sub-agents \|
32	\| 04 \| [Cyber Risk Instruction](prompts/04_cyber_risk_instruction.md) \| Security boundaries between authorized and prohibited actions \|
33
34	### Orchestration
35
36	\| # \| Pattern \| Description \|
37	\|---\|---------\|-------------\|
38	\| 05 \| [Coordinator System Prompt](prompts/05_coordinator_system_prompt.md) \| Multi-worker orchestration with phased workflows \|
39	\| 06 \| [Teammate Prompt Addendum](prompts/06_teammate_prompt_addendum.md) \| Communication protocols for multi-agent collaboration \|
40
41	### Specialized Agents
42
43	\| # \| Pattern \| Description \|
44	\|---\|---------\|-------------\|
45	\| 07 \| [Verification Agent](prompts/07_verification_agent.md) \| Adversarial testing agent that validates implementations \|
46	\| 08 \| [Explore Agent](prompts/08_explore_agent.md) \| Read-only codebase exploration with no-modify constraints \|
47	\| 09 \| [Agent Creation Architect](prompts/09_agent_creation_architect.md) \| Generates new agent configurations from requirements \|
48	\| 10 \| [Status Line Setup Agent](prompts/10_statusline_setup_agent.md) \| Terminal status line configuration across shells \|
49
50	### Security and Permissions
51
52	\| # \| Pattern \| Description \|
53	\|---\|---------\|-------------\|
54	\| 11 \| [Permission Explainer](prompts/11_permission_explainer.md) \| Risk assessment before tool approval \|
55	\| 12 \| [Auto Mode Classifier](prompts/12_yolo_auto_mode_classifier.md) \| Multi-stage security classifier for autonomous tool execution \|
56
57	### Tool Descriptions
58
59	\| # \| Pattern \| Description \|
60	\|---\|---------\|-------------\|
61	\| 13 \| [Tool-Specific Prompts](prompts/13_tool_prompts.md) \| How individual tools (Bash, Edit, Agent, etc.) describe themselves \|
62
63	### Utility Patterns
64
65	\| # \| Pattern \| Description \|
66	\|---\|---------\|-------------\|
67	\| 14 \| [Tool Use Summary](prompts/14_tool_use_summary.md) \| Generating concise labels for completed tool batches \|
68	\| 15 \| [Session Search](prompts/15_session_search.md) \| Semantic search across past conversation sessions \|
69	\| 16 \| [Memory Selection](prompts/16_memory_selection.md) \| Selecting relevant memory files for query context \|
70	\| 17 \| [Auto Mode Critique](prompts/17_auto_mode_critique.md) \| Reviewing user-written classifier rules \|
71	\| 20 \| [Session Title](prompts/20_session_title.md) \| Lightweight title generation for session management \|
72	\| 2

People who liked this also viewed

6 picks

Type