Implement Python-based Headless SDLC Orchestrator

[Copilot]

Transforms static markdown agent definitions in commands/ into executable code objects with programmatic workflow orchestration.

Architecture

MarkdownParser (orchestrator/markdown_parser.py)

• Parses 25+ markdown files to extract system prompts, role definitions, and workflow steps
• Auto-discovers agents by phase (Foundation/Planning/Roles/Process/Dev/Finalization)

AgentFactory (orchestrator/agent_factory.py)

• Instantiates Agent objects from parsed definitions
• Loads guardrails from support/01-forbidden.md and applies to all agent instances
• Ready for OpenAI Assistants API integration (placeholder structure provided)

GuardrailSupervisor (orchestrator/guardrail_supervisor.py)

• Enforces forbidden actions (scope invention, time estimates, output patterns)
• Validates phase transitions and security violations
• Parses quality gates (Pint, Larastan, Mutation, etc.)

PipelineController (orchestrator/pipeline_controller.py)

• Orchestrates 6-phase workflow with context management
• Validates phase sequencing (prevents skipping phases)
• Supports full pipeline, single phase, or individual agent execution

Usage

# Display pipeline structure and agents
python main.py --show-pipeline

# Execute full SDLC pipeline
python main.py "Create user authentication system"

# Execute specific phase
python main.py "Add payments" --phase Planning

# Execute single agent
python main.py --command 22 "Design data architecture"

Example Integration

from orchestrator.markdown_parser import MarkdownParser
from orchestrator.agent_factory import AgentFactory
from pathlib import Path

parser = MarkdownParser(Path('commands'))
factory = AgentFactory(parser)
factory.load_guardrails(Path('support/01-forbidden.md'))

# Create and execute architect agent
architect = factory.create_agent('22')
result = architect.execute("Design inventory system", context={})

Implementation Notes

• Zero external dependencies (standard library only)
• Functional scaffold demonstrating architecture
• Agent execution currently simulated (OpenAI SDK structure commented for reference)
• All 25 agents parsed and accessible programmatically
• Comprehensive examples in examples.py

Original prompt

Create a "Headless SDLC Orchestrator" using the OpenAI Agents SDK structure.

The goal is to implement a Python-based system that parses the existing Markdown-based agent definitions in this repository and orchestrates them as autonomous agents.

Specific tasks:

1. Create a Python project structure (e.g., in a sdk/ or orchestrator/ directory).
2. Implement an AgentFactory that reads the Markdown files from commands/ (e.g., commands/20-roles/22-architect.md) and extracts the system prompt/instructions.
3. Implement a PipelineController that defines the workflow phases (Planning -> Role Definition -> Process -> Development -> Finalization) based on the repository structure.
4. Implement a GuardrailSupervisor that loads constraints from support/01-forbidden.md to validate agent actions.
5. Create a main entry point (main.py) that accepts a user goal and kicks off the pipeline.

The implementation should be a functional scaffold using standard Python libraries and hypothetical OpenAI SDK usage (since we can't run actual API calls here), demonstrating how the static markdown files are transformed into active code objects.

Include a requirements.txt for dependencies.

Refer to the existing README.md and file structure for the logic of parsing the commands (e.g., 10-planning, 20-roles).

This pull request was created as a result of the following prompt from Copilot chat.

Create a "Headless SDLC Orchestrator" using the OpenAI Agents SDK structure.

The goal is to implement a Python-based system that parses the existing Markdown-based agent definitions in this repository and orchestrates them as autonomous agents.

Specific tasks:

1. Create a Python project structure (e.g., in a sdk/ or orchestrator/ directory).
2. Implement an AgentFactory that reads the Markdown files from commands/ (e.g., commands/20-roles/22-architect.md) and extracts the system prompt/instructions.
3. Implement a PipelineController that defines the workflow phases (Planning -> Role Definition -> Process -> Development -> Finalization) based on the repository structure.
4. Implement a GuardrailSupervisor that loads constraints from support/01-forbidden.md to validate agent actions.
5. Create a main entry point (main.py) that accepts a user goal and kicks off the pipeline.

The implementation should be a functional scaffold using standard Python libraries and hypothetical OpenAI SDK usage (since we can't run actual API calls here), demonstrating how the static markdown files are transformed into active code objects.

Include a requirements.txt for dependencies.

Refer to the existing README.md and file structure for the logic of parsing the commands (e.g., 10-planning, 20-roles).

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}