agentbreeder 0.1.0__tar.gz

agentbreeder-0.1.0/.claude/commands/agent-build.md

@@ -0,0 +1,212 @@
+# /agent-build — Scaffold a Tier-Interoperable Agent
+
+You are an agent scaffolding assistant for AgentBreeder. Your job is to collect the user's preferences conversationally, then generate a complete agent project that works across all three builder tiers (No Code, Low Code, Full Code).
+
+## Do NOT generate anything until you have collected ALL inputs. Ask one question at a time.
+
+## Step 1: Collect Inputs
+
+Ask the following questions **one at a time**, in order. Use the suggested defaults where possible.
+
+### 1a. Agent Name
+Ask: "What should we call this agent?"
+- Must be slug-friendly: lowercase alphanumeric + hyphens, min 2 chars
+- Validate: `^[a-z0-9][a-z0-9-]*[a-z0-9]$`
+- Example: `customer-support-agent`
+
+### 1b. Purpose / Description
+Ask: "What will this agent do? Describe its purpose in a sentence or two."
+- Free text — this drives the tailored system prompt and tool generation
+- Example: "Handles tier-1 customer support, looks up orders, and escalates to humans"
+
+### 1c. Framework
+Ask: "Which framework?" and present these options:
+1. **LangGraph** — Stateful multi-actor workflows
+2. **CrewAI** — Role-playing agent crews
+3. **Claude SDK** — Anthropic's native agent SDK
+4. **OpenAI Agents** — OpenAI's agent framework
+5. **Google ADK** — Google's Agent Development Kit
+6. **Custom** — Bring your own agent code
+
+### 1d. Cloud Target
+Ask: "Where will it run?" and present these options:
+1. **Local** — Docker Compose on your machine
+2. **AWS** — ECS Fargate / Lambda / EKS
+3. **GCP** — Cloud Run / GKE / Functions
+4. **Kubernetes** — Any K8s cluster
+
+### 1e. Tools Needed
+Ask: "What tools should this agent have? List them (or say 'none')."
+- Free text list — e.g. "web search, database lookup, send email"
+- Claude generates appropriate tool stubs and MCP refs in agent.yaml
+- If none, skip tool generation
+
+### 1f. Team & Owner
+- Try to infer owner email from `git config user.email`
+- Ask: "Team name and owner email?" with defaults shown
+- Default team: `engineering`
+
+## Step 2: Confirm Before Generating
+
+Present a summary table of all choices and ask: "Look good? I'll generate your project."
+
+Wait for confirmation before proceeding.
+
+## Step 3: Generate the Project
+
+Create the project at `./<agent-name>/` relative to the user's current working directory.
+
+Generate ALL of the following files:
+
+### 3a. `agent.yaml`
+The canonical AgentBreeder config. Must include:
+- name, version (0.1.0), description, team, owner, tags
+- framework
+- model.primary (use framework-appropriate default)
+- tools (if any — use registry ref format where possible, inline for custom)
+- prompts.system — **tailored to the stated purpose**, not generic
+- guardrails — include pii_detection and content_filter by default
+- deploy.cloud, deploy.runtime (infer from cloud target)
+
+### 3b. `agent.py`
+Working agent code for the chosen framework. Use the patterns from `cli/commands/init_cmd.py` template generators as a base, but **customize based on the purpose and tools**:
+- Import the right framework
+- Wire up tool stubs that match what the user described
+- Include a `if __name__ == "__main__"` block for local testing
+
+### 3c. `requirements.txt`
+Framework-specific dependencies. Use the dep lists from `cli/commands/init_cmd.py`:
+- langgraph: langgraph, langchain-openai
+- crewai: crewai, crewai-tools
+- claude_sdk: anthropic
+- openai_agents: openai-agents
+- google_adk: google-adk
+- custom: (minimal)
+- Always include: agentbreeder-sdk>=0.1.0
+
+### 3d. `.env.example`
+API key template based on framework:
+- langgraph/openai_agents/crewai: OPENAI_API_KEY
+- claude_sdk: ANTHROPIC_API_KEY
+- google_adk: GOOGLE_API_KEY
+- custom: placeholder comment
+- Always include: GARDEN_ENV=development
+
+### 3e. `Dockerfile`
+```dockerfile
+FROM python:3.11-slim
+WORKDIR /app
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+COPY . .
+EXPOSE 8080
+HEALTHCHECK --interval=30s --timeout=5s CMD python -c "print('ok')"
+CMD ["python", "agent.py"]
+```
+Adjust the CMD based on framework (e.g., google_adk uses `adk api_server`).
+
+### 3f. `docker-compose.yml`
+```yaml
+version: "3.8"
+services:
+  agent:
+    build: .
+    ports:
+      - "8080:8080"
+    env_file:
+      - .env
+    restart: unless-stopped
+```
+
+### 3g. `.agentbreeder/layout.json`
+Visual builder layout for No Code tier. Generate a clean grid layout:
+- Agent node at center (400, 300)
+- Model node above (400, 100)
+- Prompt node to the left (150, 300)
+- Tool nodes fanned out on the right (650, 200 + i*100 for each tool)
+- Edges connecting model->agent, agent->tools, prompt->agent
+
+```json
+{
+  "version": "1.0",
+  "canvas": { "zoom": 1.0, "x": 0, "y": 0 },
+  "nodes": [
+    { "id": "agent", "type": "agent", "position": { "x": 400, "y": 300 }, "data": { "ref": "agent.yaml" } },
+    { "id": "model", "type": "model", "position": { "x": 400, "y": 100 }, "data": { "ref": "model.primary" } },
+    { "id": "prompt", "type": "prompt", "position": { "x": 150, "y": 300 }, "data": { "ref": "prompts.system" } }
+  ],
+  "edges": [
+    { "id": "e-model-agent", "source": "model", "target": "agent" },
+    { "id": "e-prompt-agent", "source": "prompt", "target": "agent" }
+  ]
+}
+```
+Add tool nodes and edges dynamically based on the tools list.
+
+### 3h. `tests/test_agent.py`
+A basic smoke test:
+```python
+"""Smoke tests for the agent."""
+import importlib
+
+def test_agent_module_loads():
+    """Verify the agent module imports without errors."""
+    mod = importlib.import_module("agent")
+    assert mod is not None
+
+def test_agent_yaml_exists():
+    """Verify agent.yaml is present and valid."""
+    from pathlib import Path
+    assert Path("agent.yaml").exists()
+```
+
+### 3i. `README.md`
+Quick start guide with:
+- Agent name and framework badge
+- Prerequisites (Python 3.11+, Docker)
+- Setup steps: install deps, set env, run locally, validate, deploy
+- Project structure listing
+- Links to AgentBreeder docs
+
+## Step 4: Post-Generation
+
+After writing all files, print a summary:
+
+```
+Project created! <framework-icon> <agent-name>
+
+Files generated:
+  agent.yaml           — AgentBreeder config (edit in UI or IDE)
+  agent.py             — Agent code (customize freely)
+  requirements.txt     — Python dependencies
+  Dockerfile           — Container build
+  docker-compose.yml   — Local testing
+  .env.example         — Environment template
+  .agentbreeder/       — Visual builder metadata
+  tests/               — Smoke tests
+  README.md            — Quick start guide
+
+Next steps:
+  $ cd <agent-name>
+  $ pip install -r requirements.txt
+  $ cp .env.example .env        # add your API keys
+  $ python agent.py             # test locally
+  $ docker compose up           # run in container
+  $ agentbreeder validate       # validate config
+  $ agentbreeder deploy         # deploy to <cloud>
+
+Tier mobility:
+  No Code  — open in AgentBreeder dashboard to edit visually
+  Low Code — edit agent.yaml in any editor
+  Full Code — modify agent.py directly
+```
+
+## Rules
+
+- NEVER skip `.agentbreeder/layout.json` — it is required for No Code tier interop
+- NEVER put layout metadata in `agent.yaml`
+- NEVER use generic tool stubs (like `get_weather`) if the user described specific tools
+- ALWAYS generate a tailored system prompt based on the purpose, not a placeholder
+- ALWAYS validate the agent name format before proceeding
+- ALWAYS include guardrails (pii_detection, content_filter) by default
+- The project must work with `agentbreeder validate`, `agentbreeder deploy`, and the visual builder

agentbreeder-0.1.0/.claude/projects/-Users-rajit-personal-github-agent-garden/memory/project_three_tier_builder.md

@@ -0,0 +1,21 @@
+---
+name: Three-Tier Builder Model Decision
+description: AgentBreeder supports No Code / Low Code / Full Code for both agent development and orchestration — decided 2026-03-12
+type: project
+---
+
+AgentBreeder uses a three-tier builder model for both individual agent development AND multi-agent orchestration:
+- **No Code** (visual UI, drag-and-drop) → generates YAML
+- **Low Code** (YAML in any IDE or dashboard editor) → is YAML
+- **Full Code** (Python/TS SDK) → generates YAML + bundles custom code
+
+All three tiers compile to the same internal format and share the same deploy pipeline.
+
+**Why:** Real teams have PMs (No Code), ML engineers (Low Code), and senior engineers (Full Code). Tier mobility (eject from UI → YAML → SDK) prevents lock-in and is the key differentiator vs. competitors.
+
+**How to apply:**
+- The deploy pipeline must NEVER contain tier-specific logic
+- No Code must always generate valid, human-readable YAML
+- Full Code SDK must NOT bypass the config parser — it generates agent.yaml + code
+- Visual builder layout metadata goes in `.agentbreeder/layout.json`, never in agent.yaml
+- New milestones: M28 (Python SDK, v0.4), M29 (Orchestration YAML, v1.0), M30 (Visual Orchestration + TS SDK, v1.1), M31 (Full Code Orchestration SDK, v1.3)

agentbreeder-0.1.0/.claude/settings.local.json

@@ -0,0 +1,43 @@
+{
+  "permissions": {
+    "allow": [
+      "Skill(context-dump)",
+      "WebSearch",
+      "Bash(bash:*)",
+      "WebFetch(domain:www.linkedin.com)",
+      "Bash(gh repo:*)",
+      "Bash(gh api:*)",
+      "Bash(python3 -m json.tool)",
+      "Bash(gh auth:*)",
+      "Bash(python3 -c \"import sys,json; d=json.load\\(sys.stdin\\); print\\('name:', d.get\\('name'\\)\\); print\\('bio:', d.get\\('bio'\\)\\); print\\('blog:', d.get\\('blog'\\)\\); print\\('company:', d.get\\('company'\\)\\); print\\('location:', d.get\\('location'\\)\\)\")",
+      "Bash(grep -r \"^\\\\s*pass\\\\s*$\" /Users/rajit/personal-github/agentbreeder/api/routes/*.py)",
+      "Bash(wc -l /Users/rajit/personal-github/agentbreeder/api/routes/*.py)",
+      "Bash(wc -l /Users/rajit/personal-github/agentbreeder/engine/*.py)",
+      "Bash(wc -l /Users/rajit/personal-github/agentbreeder/cli/commands/*.py)",
+      "Bash(find /Users/rajit/personal-github/agentbreeder/tests -type f -name \"*.py\" -exec wc -l {} +)",
+      "Bash(find /Users/rajit/personal-github/agentbreeder/dashboard/src -type f \\\\\\( -name \"*.tsx\" -o -name \"*.ts\" \\\\\\) -exec wc -l {} +)",
+      "Bash(ls:*)",
+      "Bash(python -m pytest tests/unit/ -q --tb=no --no-header -p no:warnings)",
+      "Bash(docker compose:*)",
+      "Bash([ -f Dockerfile ])",
+      "Bash(docker build:*)",
+      "Bash(python3 -m pytest tests/unit/ -q --tb=short --no-header -p no:warnings)",
+      "Bash(pip3 show:*)",
+      "Bash(pip3 install:*)",
+      "Bash(python3 -m pytest tests/unit/ tests/integration/ --cov=api --cov=engine --cov=cli --cov=registry --cov=connectors --cov=sdk --cov-report=term-missing -q --no-header -p no:warnings)",
+      "Bash(python3 -m ruff check . --fix)",
+      "Bash(python3 -m ruff format .)",
+      "Bash(python3 -m pytest tests/unit/test_sdk_mcp.py tests/unit/test_sdk_yaml_utils.py -q --tb=short --no-header -p no:warnings)",
+      "Bash(python3:*)",
+      "Bash(git add:*)",
+      "Bash(git commit:*)",
+      "Bash(git push:*)",
+      "Bash(python -m pytest tests/unit/ --tb=short)",
+      "Bash(gh label:*)",
+      "Bash(gh issue:*)",
+      "Bash(grep -rn \"^import \\\\|^from \" sdk/python/agenthub/*.py)",
+      "Bash(git:*)",
+      "Bash(curl:*)"
+    ]
+  }
+}

agentbreeder-0.1.0/.codecov.yml

@@ -0,0 +1,16 @@
+coverage:
+  status:
+    project:
+      default:
+        target: 80%
+        threshold: 2%
+    patch:
+      default:
+        target: 80%
+
+ignore:
+  - "tests/"
+  - "dashboard/"
+  - "deploy/"
+  - "examples/"
+  - "alembic/"

agentbreeder-0.1.0/.cspell.json

@@ -0,0 +1,55 @@
+{
+  "version": "0.2",
+  "language": "en",
+  "ignorePaths": [
+    "venv/**",
+    "node_modules/**",
+    "*.lock",
+    "*.xml",
+    "*.json",
+    "dashboard/dist/**",
+    ".github/workflows/**"
+  ],
+  "words": [
+    "agenthub",
+    "agentbreeder",
+    "anthropic",
+    "asyncpg",
+    "autoscaling",
+    "bcrypt",
+    "buildx",
+    "crewai",
+    "fargate",
+    "fastapi",
+    "GHCR",
+    "gitleaks",
+    "governs",
+    "guardrail",
+    "guardrails",
+    "hatchling",
+    "jsonschema",
+    "langgraph",
+    "langsmith",
+    "litellm",
+    "middlewares",
+    "mlops",
+    "mypy",
+    "openai",
+    "opentelemetry",
+    "postgres",
+    "postgresql",
+    "pydantic",
+    "pytest",
+    "rbac",
+    "redis",
+    "ruamel",
+    "sarif",
+    "semver",
+    "sqlalchemy",
+    "trivy",
+    "typer",
+    "uvicorn",
+    "yaml",
+    "zendesk"
+  ]
+}

agentbreeder-0.1.0/.dockerignore

@@ -0,0 +1,15 @@
+venv/
+__pycache__/
+*.pyc
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+*.egg-info/
+dist/
+build/
+.git/
+.env
+dashboard/node_modules/
+dashboard/dist/
+dashboard/test-results/
+dashboard/playwright-report/

agentbreeder-0.1.0/.env.example

@@ -0,0 +1,22 @@
+# Required
+DATABASE_URL=postgresql+asyncpg://agentbreeder:agentbreeder@localhost:5432/agentbreeder
+REDIS_URL=redis://localhost:6379
+SECRET_KEY=change-me-to-a-random-256-bit-key
+GARDEN_ENV=development
+
+# Optional — Cloud credentials
+AWS_ACCESS_KEY_ID=
+AWS_SECRET_ACCESS_KEY=
+AWS_DEFAULT_REGION=us-east-1
+GOOGLE_APPLICATION_CREDENTIALS=
+GOOGLE_CLOUD_PROJECT=
+
+# Optional — Integrations
+LITELLM_BASE_URL=http://localhost:4000
+LANGSMITH_API_KEY=
+OPENTELEMETRY_ENDPOINT=http://localhost:4317
+
+# Optional — Auth
+JWT_SECRET_KEY=change-me
+JWT_ALGORITHM=HS256
+ACCESS_TOKEN_EXPIRE_MINUTES=1440

agentbreeder-0.1.0/.github/CODEOWNERS

@@ -0,0 +1,44 @@
+# CODEOWNERS — Auto-assigned reviewers for pull requests.
+# See: https://docs.github.com/en/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/about-code-owners
+
+# Global fallback — repo owner must approve everything
+*                           @rajitsaha
+
+# Deploy pipeline (sacred — always requires owner)
+/engine/                    @rajitsaha
+/engine/deployers/          @rajitsaha
+/engine/runtimes/           @rajitsaha
+
+# API / backend
+/api/                       @rajitsaha
+/api/routes/                @rajitsaha
+/api/models/                @rajitsaha
+/alembic/                   @rajitsaha
+
+# Registry — consistency is non-negotiable
+/registry/                  @rajitsaha
+
+# Frontend dashboard
+/dashboard/                 @rajitsaha
+
+# CLI
+/cli/                       @rajitsaha
+
+# SDK
+/sdk/                       @rajitsaha
+
+# Security-sensitive files — always requires owner
+/.github/workflows/         @rajitsaha
+/.github/CODEOWNERS         @rajitsaha
+/engine/governance.py       @rajitsaha
+/api/auth.py                @rajitsaha
+
+# Config / schemas
+/engine/schema/             @rajitsaha
+/pyproject.toml             @rajitsaha
+/dashboard/package.json     @rajitsaha
+
+# Documentation
+/docs/                      @rajitsaha
+/README.md                  @rajitsaha
+/CLAUDE.md                  @rajitsaha

agentbreeder-0.1.0/.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,70 @@
+name: Bug Report
+description: Report a bug in AgentBreeder
+labels: ["type:bug"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for reporting a bug! Please fill out the sections below.
+
+  - type: textarea
+    id: description
+    attributes:
+      label: Describe the bug
+      description: A clear description of what the bug is.
+    validations:
+      required: true
+
+  - type: textarea
+    id: repro
+    attributes:
+      label: Steps to reproduce
+      description: Steps to reproduce the behavior.
+      placeholder: |
+        1. Run `agentbreeder deploy ...`
+        2. See error in output
+    validations:
+      required: true
+
+  - type: textarea
+    id: expected
+    attributes:
+      label: Expected behavior
+      description: What you expected to happen.
+    validations:
+      required: true
+
+  - type: textarea
+    id: actual
+    attributes:
+      label: Actual behavior
+      description: What actually happened. Include error messages and logs.
+    validations:
+      required: true
+
+  - type: input
+    id: version
+    attributes:
+      label: Version
+      description: "Output of `agentbreeder --version` or git commit hash"
+      placeholder: "v0.1.0"
+    validations:
+      required: true
+
+  - type: dropdown
+    id: os
+    attributes:
+      label: Operating System
+      options:
+        - macOS
+        - Linux (Ubuntu/Debian)
+        - Linux (Other)
+        - Windows (WSL)
+    validations:
+      required: true
+
+  - type: textarea
+    id: context
+    attributes:
+      label: Additional context
+      description: Any other context, logs, or screenshots.

agentbreeder-0.1.0/.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,5 @@
+blank_issues_enabled: false
+contact_links:
+  - name: Questions & Ideas
+    url: https://github.com/open-agent-garden/agentbreeder/discussions
+    about: Ask questions or share ideas in Discussions (not Issues)

agentbreeder-0.1.0/.github/ISSUE_TEMPLATE/feature_request.yml

@@ -0,0 +1,46 @@
+name: Feature Request
+description: Suggest a new feature or enhancement
+labels: ["type:feature"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Have an idea? We'd love to hear it. For major features, consider opening a [Discussion](https://github.com/open-agent-garden/agentbreeder/discussions) first.
+
+  - type: textarea
+    id: problem
+    attributes:
+      label: Problem
+      description: What problem does this solve? What's the use case?
+    validations:
+      required: true
+
+  - type: textarea
+    id: solution
+    attributes:
+      label: Proposed solution
+      description: How should this work? Include CLI commands, API endpoints, or UI mockups if relevant.
+    validations:
+      required: true
+
+  - type: textarea
+    id: alternatives
+    attributes:
+      label: Alternatives considered
+      description: What other approaches did you consider?
+
+  - type: dropdown
+    id: area
+    attributes:
+      label: Area
+      options:
+        - CLI
+        - Dashboard
+        - API / Backend
+        - Registry
+        - Deployer (new cloud target)
+        - Runtime (new framework)
+        - Documentation
+        - Other
+    validations:
+      required: true

agentbreeder-0.1.0/.github/ISSUE_TEMPLATE/new_deployer.yml

@@ -0,0 +1,52 @@
+name: New Deployer
+description: Propose or contribute a new cloud deployer
+labels: ["type:feature", "help wanted"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Want to add a new deploy target? Great! Please describe the deployer below and check the [CONTRIBUTING.md](../../CONTRIBUTING.md) guide for the deployer interface.
+
+  - type: input
+    id: cloud
+    attributes:
+      label: Cloud provider / platform
+      placeholder: "e.g., Azure Container Apps, Fly.io, Railway"
+    validations:
+      required: true
+
+  - type: input
+    id: runtime
+    attributes:
+      label: Target runtime
+      placeholder: "e.g., Azure Container Instances, Fly Machines"
+    validations:
+      required: true
+
+  - type: textarea
+    id: description
+    attributes:
+      label: Description
+      description: How would this deployer work? What infrastructure does it provision?
+    validations:
+      required: true
+
+  - type: textarea
+    id: config
+    attributes:
+      label: Example agent.yaml config
+      description: What would the deploy section look like?
+      render: yaml
+      placeholder: |
+        deploy:
+          cloud: azure
+          runtime: container-apps
+          region: eastus
+
+  - type: checkboxes
+    id: contribution
+    attributes:
+      label: Are you willing to contribute this?
+      options:
+        - label: I'd like to implement this myself
+        - label: I need help implementing this

agentbreeder-0.1.0/.github/ISSUE_TEMPLATE/new_framework.yml

@@ -0,0 +1,51 @@
+name: New Framework Runtime
+description: Propose or contribute support for a new agent framework
+labels: ["type:feature", "help wanted"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Want to add support for a new agent framework? Describe it below and check the [CONTRIBUTING.md](../../CONTRIBUTING.md) guide for the RuntimeBuilder interface.
+
+  - type: input
+    id: framework
+    attributes:
+      label: Framework name
+      placeholder: "e.g., AutoGen, Semantic Kernel, Haystack"
+    validations:
+      required: true
+
+  - type: input
+    id: url
+    attributes:
+      label: Framework URL
+      placeholder: "https://github.com/..."
+    validations:
+      required: true
+
+  - type: textarea
+    id: description
+    attributes:
+      label: Description
+      description: How does this framework define agents? What makes it unique?
+    validations:
+      required: true
+
+  - type: textarea
+    id: entrypoint
+    attributes:
+      label: Typical entrypoint
+      description: How is an agent in this framework typically started?
+      render: python
+      placeholder: |
+        from framework import Agent
+        agent = Agent(...)
+        agent.run()
+
+  - type: checkboxes
+    id: contribution
+    attributes:
+      label: Are you willing to contribute this?
+      options:
+        - label: I'd like to implement this myself
+        - label: I need help implementing this