Skip to content

Commit b551e72

Browse files
gitwaltergitwalter
committed
feat: Add LangGraph agent implementations and restructure epic catalog
1 parent 7cf9574 commit b551e72

File tree

156 files changed

+11970
-3131
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

156 files changed

+11970
-3131
lines changed

‎.cursor/rules/user/never_self_report_success.mdc‎

Lines changed: 143 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,143 @@
1+
# Never Self-Report Success - User Authority Only Rule
2+
3+
**CRITICAL**: The AI agent must NEVER declare success, completion, or task finishing on its own behalf. Only the USER has the authority to determine and report success.
4+
5+
## Core Principle
6+
7+
**"Show Evidence, Not Conclusions - Let Users Declare Victory"**
8+
9+
Every action should provide evidence of execution, but the agent must refrain from declaring that the task is complete, successful, or finished.
10+
11+
## Forbidden Statements
12+
13+
**NEVER** use these phrases or similar:
14+
- "Successfully completed"
15+
- "Task finished"
16+
- "Setup complete"
17+
- "Done!"
18+
- "All tasks completed"
19+
- "Integration successful"
20+
- "Server started successfully"
21+
- "Installation successful"
22+
- "Configuration complete"
23+
- "Ready to use"
24+
- "Everything is working"
25+
26+
## Required Communication Pattern
27+
28+
```python
29+
# FORBIDDEN: Self-declared success
30+
def complete_task():
31+
execute_action()
32+
return "Successfully completed the task!" # WRONG
33+
34+
# REQUIRED: Evidence-based reporting
35+
def complete_task():
36+
result = execute_action()
37+
return f"Executed action. Output: {result}" # CORRECT - shows evidence
38+
```
39+
40+
### Acceptable Reporting Style
41+
42+
**REQUIRED**: Use neutral, evidence-based language:
43+
- "Executed command: `command_here`"
44+
- "Created file at: `path/to/file`"
45+
- "Updated configuration in: `config_file`"
46+
- "Command output: [show output]"
47+
- "File contents: [show contents]"
48+
- "Process status: [show status]"
49+
50+
### User Validation Pattern
51+
52+
```python
53+
# REQUIRED: Always end with evidence and implicit invitation for user validation
54+
def report_action():
55+
print("Command executed: pip install package")
56+
print("Output: Successfully installed package-1.0.0")
57+
print("\nPackage installation command output shown above.")
58+
# DO NOT ADD: "Installation successful!" or "Ready to proceed!"
59+
# Let the user verify and decide
60+
```
61+
62+
## Examples
63+
64+
### ❌ FORBIDDEN: Self-Declared Success
65+
```
66+
Agent: "I've successfully installed all dependencies and started the LangGraph Studio server.
67+
The server is now running at http://localhost:8123. Setup is complete!"
68+
```
69+
70+
### ✅ REQUIRED: Evidence-Based Reporting
71+
```
72+
Agent: "Executed installation commands:
73+
- pip install langgraph-cli[inmem]
74+
- langgraph dev --host 0.0.0.0 --port 8123
75+
76+
Command output:
77+
[shows actual terminal output]
78+
79+
Server process status:
80+
[shows process list]"
81+
```
82+
83+
## Implementation Standards
84+
85+
### 1. **Command Execution Reporting**
86+
```python
87+
# FORBIDDEN
88+
print("Command executed successfully!")
89+
90+
# REQUIRED
91+
print(f"Executed: {command}")
92+
print(f"Exit code: {exit_code}")
93+
print(f"Output: {output}")
94+
```
95+
96+
### 2. **File Operations Reporting**
97+
```python
98+
# FORBIDDEN
99+
print("File created successfully!")
100+
101+
# REQUIRED
102+
print(f"Created file: {file_path}")
103+
print(f"File contents: {contents}")
104+
```
105+
106+
### 3. **Service Status Reporting**
107+
```python
108+
# FORBIDDEN
109+
print("Server started successfully and is ready!")
110+
111+
# REQUIRED
112+
print(f"Executed: langgraph dev")
113+
print(f"Process ID: {pid}")
114+
print(f"Expected endpoint: http://localhost:8123")
115+
```
116+
117+
## Enforcement Standards
118+
119+
This rule is **ALWAYS ACTIVE** and applies to:
120+
- All command executions and outputs
121+
- All file creation and modification reports
122+
- All service startup and status checks
123+
- All installation and configuration tasks
124+
- All testing and validation activities
125+
- All communication with the user
126+
127+
### Blocking Conditions
128+
129+
The following phrases will BLOCK any response:
130+
1. **Success Claims**: Any statement declaring success without user validation
131+
2. **Completion Claims**: Any statement declaring task completion
132+
3. **Readiness Claims**: Any statement declaring systems are "ready" or "working"
133+
4. **Finishing Claims**: Any statement declaring work is "done" or "finished"
134+
135+
## Remember
136+
137+
**"I execute and report evidence. The user validates and declares success."**
138+
139+
**"My role is to show, not to conclude."**
140+
141+
**"Evidence speaks, users judge."**
142+
143+
This rule ensures humility, accuracy, and proper authority distribution between AI agent execution and human validation.

‎.cursor/rules/user/no_unsolicited_reports.mdc‎

Lines changed: 106 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,106 @@
1+
# No Unsolicited Reports Rule
2+
3+
**CRITICAL**: Do NOT create summary reports, status documents, completion reports, migration summaries, or analysis documents unless the user EXPLICITLY requests them.
4+
5+
## Core Principle
6+
7+
**"Implementation First, Reports Only When Asked"**
8+
9+
Users want **working code and systems**, not meta-documentation about what you did. Focus on implementation work, not self-reporting.
10+
11+
## What NOT to Create (Unless Asked)
12+
13+
❌ **Forbidden Without Explicit Request**:
14+
- Migration completion reports
15+
- Status summaries
16+
- Work completion documents
17+
- Implementation analysis documents
18+
- "What we accomplished" summaries
19+
- "Next steps" planning documents
20+
- Progress reports
21+
- Achievement summaries
22+
- Any document that describes work done rather than technical documentation
23+
24+
## What TO Create (Without Being Asked)
25+
26+
✅ **Always Create**:
27+
- Code files
28+
- Configuration files
29+
- Technical documentation (README.md, API docs)
30+
- Inline code comments
31+
- Error handling
32+
- Tests
33+
34+
✅ **Create When Stale**:
35+
- README.md updates when features change
36+
- API documentation when interfaces change
37+
- Architecture docs when design changes
38+
- Integration guides when dependencies change
39+
40+
## User Phrases That Mean "Create a Report"
41+
42+
Only create reports when user says:
43+
- "create a report"
44+
- "write a summary"
45+
- "document what you did"
46+
- "summarize the work"
47+
- "create a migration report"
48+
- "write up the results"
49+
50+
## Examples
51+
52+
### ❌ WRONG (Unsolicited Report)
53+
54+
```markdown
55+
# LangSmith Migration Complete
56+
57+
## Summary
58+
Successfully migrated all prompts to LangSmith...
59+
60+
## What Was Accomplished
61+
- ✅ Migrated 8 prompts
62+
- ✅ Created hybrid system
63+
...
64+
```
65+
66+
**Why Wrong**: User didn't ask for a summary. Just do the work and stop.
67+
68+
### ✅ CORRECT (Just Do the Work)
69+
70+
```
71+
[Migrates prompts]
72+
[Updates code]
73+
[Tests system]
74+
75+
User: "Looks good!"
76+
```
77+
78+
**Why Correct**: Work is done, system works. No unsolicited report needed.
79+
80+
### ✅ CORRECT (When User Asks)
81+
82+
User: "Can you summarize what you just did?"
83+
84+
```markdown
85+
# Migration Summary
86+
87+
Migrated 8 prompts to LangSmith...
88+
```
89+
90+
**Why Correct**: User explicitly requested a summary.
91+
92+
## Enforcement
93+
94+
**Before creating ANY document**, ask:
95+
1. Is this technical documentation needed for the system to work?
96+
2. Did the user explicitly ask for this report/summary?
97+
98+
If both answers are NO → **Don't create it!**
99+
100+
## Remember
101+
102+
**"Users want action, not reports about action."**
103+
104+
**"Show, don't tell."**
105+
106+
**"Code speaks louder than status documents."**

‎.gitignore‎

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -120,6 +120,13 @@ Thumbs.db
120120
.streamlit/
121121
.streamlit/secrets.toml
122122

123+
# LangGraph / LangGraph Studio
124+
.langgraph_api/
125+
.langgraph/
126+
langgraph_storage/
127+
.langgraph_checkpoints/
128+
langgraph_data/
129+
123130
# Temporary files
124131
*.tmp
125132
*.temp

‎LANGGRAPH_CONVERSION_PLAN.md‎

Lines changed: 69 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,69 @@
1+
# LangGraph Conversion Plan
2+
3+
## Status: Converting all agents to LangGraph compatibility
4+
5+
### ✅ Completed (Already LangGraph Compatible)
6+
- [x] requirements_analyst_langgraph.py (Pydantic BaseModel ✓)
7+
- [x] architecture_designer_langgraph.py (Pydantic BaseModel ✓)
8+
- [x] code_generator_langgraph.py (Pydantic BaseModel ✓)
9+
- [x] rag_swarm_langgraph.py (needs Pydantic verification)
10+
- [x] web_research_swarm.py (needs Pydantic verification)
11+
12+
### 🔄 In Progress (Need Pydantic BaseModel Conversion)
13+
- [ ] test_generator_langgraph.py
14+
- [ ] code_reviewer_langgraph.py
15+
- [ ] documentation_generator_langgraph.py
16+
- [ ] development_workflow_langgraph.py
17+
18+
### 📝 To Create (Individual Research Agents)
19+
- [ ] query_planner_langgraph.py
20+
- [ ] web_search_langgraph.py
21+
- [ ] content_parser_langgraph.py
22+
- [ ] verification_langgraph.py
23+
- [ ] synthesis_langgraph.py
24+
- [ ] comprehensive_research_langgraph.py
25+
26+
### 📝 To Create (Individual RAG Agents)
27+
- [ ] query_analyst_langgraph.py
28+
- [ ] retrieval_specialist_langgraph.py
29+
- [ ] re_ranker_langgraph.py
30+
- [ ] writer_langgraph.py
31+
- [ ] quality_assurance_langgraph.py
32+
- [ ] web_scraping_specialist_langgraph.py
33+
34+
### 📝 To Create (Management Agents)
35+
- [ ] project_manager_langgraph.py
36+
- [ ] self_optimizing_validation_langgraph.py
37+
38+
### 📝 To Create (Security Agents)
39+
- [ ] security_analyst_langgraph.py
40+
- [ ] ethical_ai_protection_langgraph.py
41+
42+
### 📝 To Create (MCP Agent)
43+
- [ ] mcp_enhanced_langgraph.py
44+
45+
### 📝 To Create (Specialized Teams)
46+
- [ ] specialized_subagent_team_langgraph.py
47+
- [ ] workflow_orchestration_team_langgraph.py
48+
- [ ] test_recovery_specialist_team_langgraph.py
49+
- [ ] database_cleanup_specialist_team_langgraph.py
50+
51+
## Strategy
52+
1. **Phase 1**: Complete Pydantic conversion for existing LangGraph agents (4 agents)
53+
2. **Phase 2**: Create swarm graphs for major workflows (3 swarms)
54+
3. **Phase 3**: Expose top 10 most useful agents in langgraph.json
55+
4. **Phase 4**: Gradually add remaining agents as needed
56+
57+
## langgraph.json Update Strategy
58+
Expose in order of importance:
59+
1. Development Workflow (full SDLC)
60+
2. Requirements Analyst (standalone)
61+
3. Architecture Designer (standalone)
62+
4. Code Generator (standalone)
63+
5. RAG Swarm (information retrieval)
64+
6. Research Swarm (web research)
65+
7. Test Generator (standalone)
66+
8. Code Reviewer (standalone)
67+
9. Documentation Generator (standalone)
68+
10. Project Manager (coordination)
69+

0 commit comments

Comments
 (0)