API Reference

This document provides detailed API reference for Westodyssey framework.

Core Classes

TangSeng Class

The main orchestrator class that represents the human user (Tang Sanzang).

from westodyssey import TangSeng

# Initialize TangSeng
master = TangSeng(
    config_path="config/team_config.json",
    workflow_path="config/workflow.yaml"
)

Constructor Parameters

Parameter	Type	Description	Default
`config_path`	str	Path to team configuration file	`"config/team_config.json"`
`workflow_path`	str	Path to workflow configuration file	`"config/workflow.yaml"`
`api_key`	str	OpenAI API key (optional)	`None` (reads from .env)
`verbose`	bool	Enable verbose logging	`False`

Methods

`assign_mission(mission, context=None)`

Assign a mission to the team.

result = master.assign_mission(
    mission="Analyze trending AI projects",
    context="Focus on multi-agent systems"
)

Parameters:

mission (str): The main mission/task description
context (str, optional): Additional context or constraints

Returns:

dict: Mission result with status and output

`get_team_status()`

Get current status of all team members.

status = master.get_team_status()

Returns:

dict: Team status information

`reset_team()`

Reset the team to initial state.

master.reset_team()

Wukong Class

The core problem solver (Sun Wukong).

from westodyssey.agents import Wukong

wukong = Wukong(
    model="gpt-4",
    temperature=0.7,
    tools=["browser", "code_interpreter"]
)

Constructor Parameters

Parameter	Type	Description	Default
`model`	str	LLM model to use	`"gpt-4"`
`temperature`	float	Creativity level (0.0-1.0)	`0.7`
`tools`	list	Available tools	`["browser", "code_interpreter"]`
`max_tokens`	int	Maximum tokens per response	`4000`

Methods

`solve(problem, context=None)`

Solve a complex problem.

solution = wukong.solve(
    problem="Write a web scraper for GitHub trends",
    context="Use Python and BeautifulSoup"
)

`generate_code(language, task)`

Generate code for a specific task.

code = wukong.generate_code(
    language="python",
    task="Data analysis with pandas"
)

`research(topic, sources=5)`

Research a topic and gather information.

research_result = wukong.research(
    topic="AI agent frameworks",
    sources=5
)

Bajie Class

The quality auditor (Zhu Bajie).

from westodyssey.agents import Bajie

bajie = Bajie(
    model="claude-3-sonnet",
    strictness="high"
)

Constructor Parameters

Parameter	Type	Description	Default
`model`	str	LLM model to use	`"claude-3-sonnet"`
`strictness`	str	Review strictness level	`"high"`
`max_review_cycles`	int	Maximum review cycles	`3`

Methods

`review(solution, criteria=None)`

Review a solution for quality and safety.

review_result = bajie.review(
    solution=wukong_solution,
    criteria=["security", "efficiency", "best_practices"]
)

`audit_code(code, language="python")`

Perform code audit.

audit_result = bajie.audit_code(
    code=python_code,
    language="python"
)

`validate_output(output, format_spec)`

Validate output format.

validation_result = bajie.validate_output(
    output=data_output,
    format_spec={"type": "markdown", "sections": ["introduction", "analysis"]}
)

Wujing Class

The data deliverer (Sha Wujing).

from westodyssey.agents import Wujing

wujing = Wujing(
    model="gpt-3.5-turbo",
    temperature=0.0
)

Constructor Parameters

Parameter	Type	Description	Default
`model`	str	LLM model to use	`"gpt-3.5-turbo"`
`temperature`	float	Strictness level	`0.0`

Methods

`format(data, format_type="markdown")`

Format data into specified format.

formatted = wujing.format(
    data=raw_data,
    format_type="markdown"
)

`clean_data(data, rules=None)`

Clean and normalize data.

cleaned = wujing.clean_data(
    data=raw_dataset,
    rules=["remove_duplicates", "normalize_dates"]
)

`deliver(output, destination)`

Deliver output to destination.

delivery_result = wujing.deliver(
    output=final_output,
    destination="file://output/report.md"
)

Steed Class

The memory system (White Dragon Horse).

from westodyssey.agents import Steed

steed = Steed(
    vector_db_path="memory_database",
    embedding_model="text-embedding-3-small"
)

Constructor Parameters

Parameter	Type	Description	Default
`vector_db_path`	str	Path to vector database	`"memory_database"`
`embedding_model`	str	Embedding model to use	`"text-embedding-3-small"`

Methods

`record(event, metadata=None)`

Record an event to memory.

steed.record(
    event="mission_completed",
    metadata={"mission_id": "123", "duration": "1.5h"}
)

`recall(query, limit=5)`

Recall similar memories.

memories = steed.recall(
    query="data analysis mission",
    limit=5
)

`get_context(mission_id)`

Get context for a specific mission.

context = steed.get_context(mission_id="123")

Workflow API

Pilgrimage Workflow

The main workflow orchestrator.

from westodyssey.workflow import Pilgrimage

workflow = Pilgrimage(
    wukong=wukong,
    bajie=bajie, 
    wujing=wujing,
    steed=steed,
    human_in_the_loop=True
)

Methods

`dispatch(mission, context=None)`

Dispatch a mission through the full workflow.

result = workflow.dispatch(
    mission="Analyze AI trends",
    context="Latest developments in 2024"
)

`get_progress(mission_id)`

Get progress of a specific mission.

progress = workflow.get_progress(mission_id="123")

`pause_mission(mission_id)`

Pause a mission for human review.

workflow.pause_mission(mission_id="123")

Configuration API

Team Configuration

Team configuration structure:

team_config = {
    "wukong": {
        "model": "gpt-4",
        "temperature": 0.7,
        "tools": ["browser", "code_interpreter", "web_search"]
    },
    "bajie": {
        "model": "claude-3-sonnet", 
        "strictness": "high",
        "max_review_cycles": 3
    },
    "wujing": {
        "model": "gpt-3.5-turbo",
        "temperature": 0.0
    },
    "steed": {
        "embedding_model": "text-embedding-3-small",
        "vector_db_path": "memory_database"
    }
}

Workflow Configuration

Workflow configuration structure:

workflow:
  steps:
    - load_context: true
    - solve_problem: true  
    - review_solution: true
    - format_output: true
    - record_memory: true
  
  constraints:
    max_review_cycles: 3
    require_human_approval: true
    timeout_minutes: 30
  
  notifications:
    on_pause: true
    on_completion: true
    on_error: true

Utility Functions

`load_config(config_path)`

Load configuration from file.

from westodyssey.utils import load_config

config = load_config("config/team_config.json")

`save_config(config, config_path)`

Save configuration to file.

from westodyssey.utils import save_config

save_config(team_config, "config/team_config.json")

`validate_config(config)`

Validate configuration structure.

from westodyssey.utils import validate_config

is_valid = validate_config(team_config)

Error Handling

Custom Exceptions

from westodyssey.exceptions import (
    MissionTimeoutError,
    ReviewCycleExceededError,
    FormatValidationError,
    MemoryRetrievalError
)

try:
    result = workflow.dispatch(mission)
except MissionTimeoutError as e:
    print(f"Mission timed out: {e}")
except ReviewCycleExceededError as e:
    print(f"Too many review cycles: {e}")

Error Recovery

from westodyssey.recovery import retry_with_backoff

@retry_with_backoff(max_retries=3)
def execute_mission(mission):
    return workflow.dispatch(mission)

Examples

Basic Usage

from westodyssey import TangSeng

# Initialize
master = TangSeng()

# Execute mission
result = master.assign_mission(
    mission="Create a market analysis report",
    context="Focus on AI agent market in 2024"
)

print(result["output"])

Advanced Configuration

from westodyssey.workflow import Pilgrimage
from westodyssey.agents import Wukong, Bajie, Wujing, Steed

# Create custom team
wukong = Wukong(model="gpt-4o", tools=["all"])
bajie = Bajie(model="claude-3-5-sonnet", strictness="extreme")
wujing = Wujing(model="gpt-4", temperature=0.0)
steed = Steed(embedding_model="text-embedding-3-large")

# Create workflow
workflow = Pilgrimage(
    wukong=wukong,
    bajie=bajie,
    wujing=wujing,
    steed=steed,
    human_in_the_loop=False  # Full automation
)

# Execute
result = workflow.dispatch("Complex data analysis task")

Note: This API reference is continuously updated. Check the GitHub repository for the latest version.

FilesExpand file tree

api-reference.md

Latest commit

History

api-reference.md

File metadata and controls

API Reference

Core Classes

TangSeng Class

Constructor Parameters

Methods

assign_mission(mission, context=None)

get_team_status()

reset_team()

Wukong Class

Constructor Parameters

Methods

solve(problem, context=None)

generate_code(language, task)

research(topic, sources=5)

Bajie Class

Constructor Parameters

Methods

review(solution, criteria=None)

audit_code(code, language="python")

validate_output(output, format_spec)

Wujing Class

Constructor Parameters

Methods

format(data, format_type="markdown")

clean_data(data, rules=None)

deliver(output, destination)

Steed Class

Constructor Parameters

Methods

record(event, metadata=None)

recall(query, limit=5)

get_context(mission_id)

Workflow API

Pilgrimage Workflow

Methods

dispatch(mission, context=None)

get_progress(mission_id)

pause_mission(mission_id)

Configuration API

Team Configuration

Workflow Configuration

Utility Functions

load_config(config_path)

save_config(config, config_path)

validate_config(config)

Error Handling

Custom Exceptions

Error Recovery

Examples

Basic Usage

Advanced Configuration

`assign_mission(mission, context=None)`

`get_team_status()`

`reset_team()`

`solve(problem, context=None)`

`generate_code(language, task)`

`research(topic, sources=5)`

`review(solution, criteria=None)`

`audit_code(code, language="python")`

`validate_output(output, format_spec)`

`format(data, format_type="markdown")`

`clean_data(data, rules=None)`

`deliver(output, destination)`

`record(event, metadata=None)`

`recall(query, limit=5)`

`get_context(mission_id)`

`dispatch(mission, context=None)`

`get_progress(mission_id)`

`pause_mission(mission_id)`

`load_config(config_path)`

`save_config(config, config_path)`

`validate_config(config)`