AI-SDK-CREWAI

Role-based autonomous crews for delegated mission execution.

Current Features

• Agents Army routing with skill-aware primary agent selection.
• FastAPI service and local CLI entrypoint.
• CrewAI crew scaffold creation with graceful runtime-config messaging.
• Docker packaging, CI workflow, and pytest contract tests.
• Portfolio metadata, strategy notes, and skill matrix.

Existing Runtime Flow

1. A user sends a mission through the CLI or POST /run.
2. The shared Agents Army router scores the mission against agent skills.
3. The adapter selects a primary agent and support agents.
4. The CrewAI integration validates the framework-specific execution path.
5. The response returns routing, support agents, result text, and verification notes.

SDK-Specific Implementation

Builds a CrewAI Agent, Task, and Crew scaffold from the routed mission.

Repository Structure

• app.py: framework adapter and mission execution entrypoint.
• api.py: FastAPI health and run endpoints.
• runner.py: local CLI demo runner.
• core/agents_army_core/: shared routing, mission models, registry, and instruction rendering.
• tests/: contract tests for routing and verification behavior.
• .github/workflows/ci.yml: automated pytest checks.
• Dockerfile: container packaging for API deployment.
• SKILLSET.md: skills represented by this SDK adapter.

Run Locally

python3 -m pip install -r requirements.txt
python3 runner.py --mission "build secure api, add tests, and deploy"

Run As An API

uvicorn api:app --host 0.0.0.0 --port 8000

curl http://localhost:8000/health
curl -X POST http://localhost:8000/run \
  -H "content-type: application/json" \
  -d '{"mission":"secure audit and deploy an AI workflow"}'

Test

python3 -m pytest

Required Skill Set

See SKILLSET.md for the full skill matrix. At a high level this repository demonstrates:

• CrewAI integration.
• Agent routing and mission planning.
• API design with clear input/output contracts.
• CI-backed verification.
• Secure environment-based configuration.

Upgrade Roadmap

• Add specialized crew templates for engineering, security, and deployment missions.
• Connect tool permissions and execution guardrails.
• Persist task outputs and verification evidence.

Clean Repository Policy

Generated files such as __pycache__/, .pytest_cache/, local .env files, and dependency folders are ignored. Source, docs, tests, and deployment assets are intentionally kept in the repository.