npm - @bluefly/openstandardagents - Versions diffs - 0.3.3 → 0.3.6 - Mend

@bluefly/openstandardagents 0.3.3 → 0.3.6

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (1737) hide show

package/spec/v0.3.3/extensions/crewai.md DELETED Viewed

@@ -1,1575 +0,0 @@
-# CrewAI Extension for OSSA v0.3.3
-## Overview
-The `extensions.crewai` schema provides comprehensive bidirectional mapping between CrewAI's multi-agent orchestration framework and OSSA v0.3.3 specifications. This extension enables:
-- Converting CrewAI crews to OSSA Workflows with multiple agents
-- Mapping CrewAI agent roles to OSSA Agent taxonomy
-- Translating CrewAI tasks to OSSA Task kind
-- Binding CrewAI tools to OSSA capabilities
-- Supporting sequential and hierarchical process types
-- Enabling delegation patterns through OSSA DelegationConfig
-- Persisting memory state via OSSA state management
-- Integrating callbacks with OSSA observability
-## Schema Definition
-```yaml
-extensions:
-  crewai:
-    type: object
-    description: "CrewAI multi-agent orchestration framework extension"
-    properties:
-      crew_name:
-        type: string
-        description: "Name of the CrewAI crew"
-        pattern: "^[a-z][a-z0-9_-]*$"
-        examples:
-          - "research-team"
-          - "content-creation-crew"
-      agents:
-        type: array
-        description: "CrewAI agents mapped to OSSA agents"
-        items:
-          $ref: "#/definitions/CrewAIAgent"
-      tasks:
-        type: array
-        description: "CrewAI tasks mapped to OSSA workflow steps"
-        items:
-          $ref: "#/definitions/CrewAITask"
-      process_type:
-        type: string
-        enum:
-          - sequential
-          - hierarchical
-          - consensual
-        default: sequential
-        description: "Process execution pattern - maps to OSSA workflow step ordering"
-      delegation_enabled:
-        type: boolean
-        default: true
-        description: "Enable agent-to-agent delegation"
-      memory_config:
-        $ref: "#/definitions/CrewAIMemory"
-        description: "Memory system configuration"
-      manager_llm:
-        type: object
-        description: "LLM configuration for hierarchical manager (when process_type=hierarchical)"
-        properties:
-          provider:
-            type: string
-            enum: [openai, anthropic, google, azure, ollama]
-          model:
-            type: string
-        additionalProperties: true
-      max_rpm:
-        type: integer
-        minimum: 1
-        description: "Maximum requests per minute for rate limiting"
-      share_crew:
-        type: boolean
-        default: false
-        description: "Share telemetry data with CrewAI platform"
-      callbacks:
-        $ref: "#/definitions/CrewAICallbacks"
-        description: "Callback hooks for observability integration"
-      verbose:
-        type: boolean
-        default: false
-        description: "Enable verbose logging"
-      language:
-        type: string
-        default: "en"
-        description: "Primary language for agent interactions"
-```
-## Bidirectional Mapping Tables
-### CrewAI Crew to OSSA Workflow
-| CrewAI Crew Field | OSSA Workflow Field | Notes |
-|-------------------|---------------------|-------|
-| `name` | `metadata.name` | Converted to DNS-1123 format |
-| `agents` | `extensions.crewai.agents` + separate Agent manifests | Each agent becomes OSSA Agent |
-| `tasks` | `spec.steps[]` | Tasks become WorkflowStep |
-| `process` | `extensions.crewai.process_type` | Sequential/hierarchical patterns |
-| `verbose` | `spec.observability.logging.level` | Maps to debug/info |
-| `memory` | `extensions.crewai.memory_config` | Short/long-term memory |
-| `cache` | `runtime.caching` | Result caching |
-| `max_rpm` | `extensions.crewai.max_rpm` | Rate limiting |
-| `share_crew` | `extensions.crewai.share_crew` | Telemetry sharing |
-| `manager_llm` | `extensions.crewai.manager_llm` | Hierarchical manager |
-| `step_callback` | `extensions.crewai.callbacks.on_step` | Step completion hook |
-| `task_callback` | `extensions.crewai.callbacks.on_task` | Task completion hook |
-### CrewAI Agent to OSSA Agent
-| CrewAI Agent Field | OSSA Agent Field | Notes |
-|--------------------|------------------|-------|
-| `role` | `spec.role` | System prompt / role definition |
-| `goal` | `metadata.annotations["crewai.goal"]` | Agent goal |
-| `backstory` | `spec.prompts.system.template` | Combined with role |
-| `llm` | `spec.llm` | LLM configuration |
-| `tools` | `spec.tools[]` | Tool bindings |
-| `allow_delegation` | `spec.delegation.enabled` | Delegation capability |
-| `verbose` | `spec.observability.logging.level` | Logging verbosity |
-| `memory` | `spec.state.persistence` | Memory persistence |
-| `max_iter` | `spec.autonomy.max_iterations` | Iteration limit |
-| `max_rpm` | `spec.constraints.rate_limit.requests_per_minute` | Rate limit |
-| `step_callback` | `spec.observability.callbacks.on_step` | Step hook |
-### CrewAI Agent Role to OSSA Taxonomy
-| CrewAI Role Pattern | OSSA Taxonomy Domain | OSSA Agent Type | Access Tier |
-|--------------------|----------------------|-----------------|-------------|
-| `researcher`, `analyst` | `data` | `analyzer` | `tier_1_read` |
-| `writer`, `content creator` | `content` | `worker` | `tier_2_write_limited` |
-| `editor`, `reviewer` | `documentation` | `critic` | `tier_1_read` |
-| `manager`, `coordinator` | `agents` | `orchestrator` | `tier_3_write_elevated` |
-| `developer`, `coder` | `development` | `worker` | `tier_2_write_limited` |
-| `qa`, `tester` | `development` | `analyzer` | `tier_1_read` |
-| `security analyst` | `security` | `analyzer` | `tier_1_read` |
-| `devops`, `operator` | `infrastructure` | `operator` | `tier_3_write_elevated` |
-### CrewAI Task to OSSA Task/WorkflowStep
-| CrewAI Task Field | OSSA Field | Notes |
-|-------------------|------------|-------|
-| `description` | `metadata.description` | Task description |
-| `agent` | `spec.steps[].ref` | Reference to agent |
-| `expected_output` | `spec.output` | Output schema |
-| `tools` | `spec.capabilities[]` | Required capabilities |
-| `context` | `spec.steps[].input` | Input from previous tasks |
-| `async_execution` | `spec.batch.enabled` | Async pattern |
-| `callback` | `extensions.crewai.callbacks.on_task` | Task hook |
-| `output_json` | `spec.output.type: object` | JSON output |
-| `output_pydantic` | `spec.output` + Zod validation | Structured output |
-| `output_file` | `spec.output` + file capability | File output |
-| `human_input` | `spec.safety.human_approval.required` | Human-in-loop |
-### CrewAI Tools to OSSA Capabilities
-| CrewAI Tool Type | OSSA Capability | OSSA Tool Type |
-|------------------|-----------------|----------------|
-| `@tool` decorator | Custom capability | `function` |
-| `SerperDevTool` | `web_search` | `http` |
-| `ScrapeWebsiteTool` | `web_scrape` | `http` |
-| `FileReadTool` | `file_read` | `function` |
-| `DirectoryReadTool` | `directory_read` | `function` |
-| `MDXSearchTool` | `document_search` | `function` |
-| `PDFSearchTool` | `pdf_search` | `function` |
-| `DOCXSearchTool` | `docx_search` | `function` |
-| `CSVSearchTool` | `csv_search` | `function` |
-| `JSONSearchTool` | `json_search` | `function` |
-| `XMLSearchTool` | `xml_search` | `function` |
-| `CodeInterpreterTool` | `code_execute` | `function` |
-| `BrowserbaseLoadTool` | `browser_automation` | `http` |
-| `GithubSearchTool` | `github_search` | `api` |
-| `YoutubeChannelSearchTool` | `youtube_search` | `api` |
-| `YoutubeVideoSearchTool` | `youtube_video_search` | `api` |
-| `RagTool` | `rag_query` | `function` |
-| `LlamaIndexTool` | `llama_index` | `function` |
-| `SeleniumScrapingTool` | `web_scrape_dynamic` | `http` |
-### CrewAI Process Types to OSSA Workflow Patterns
-| CrewAI Process | OSSA Pattern | Description |
-|----------------|--------------|-------------|
-| `Process.sequential` | Linear `steps[]` | Tasks execute in order |
-| `Process.hierarchical` | Manager agent + `Parallel` steps | Manager delegates to workers |
-| `Process.consensual` | `Parallel` with voting step | Agents reach consensus |
-### CrewAI Memory to OSSA State
-| CrewAI Memory Type | OSSA State Field | Notes |
-|--------------------|------------------|-------|
-| `short_term` | `spec.state.context_window` | Recent interaction memory |
-| `long_term` | `spec.state.persistence.type: database` | Persistent memory |
-| `entity` | `spec.state.memory.entities` | Entity extraction |
-| `contextual` | `spec.state.memory.contextual` | RAG-based memory |
-| `embeddings` | `spec.state.memory.embeddings` | Vector store |
-### CrewAI Callbacks to OSSA Observability
-| CrewAI Callback | OSSA Observability | Notes |
-|-----------------|-------------------|-------|
-| `step_callback` | `spec.observability.tracing.on_step` | Step completion |
-| `task_callback` | `spec.observability.tracing.on_task` | Task completion |
-| `on_run_start` | `spec.observability.callbacks.before_execute` | Crew start |
-| `on_run_end` | `spec.observability.callbacks.after_execute` | Crew end |
-| `on_agent_start` | `spec.observability.callbacks.agent_start` | Agent activation |
-| `on_agent_end` | `spec.observability.callbacks.agent_end` | Agent completion |
-## Definitions
-### CrewAIAgent
-```yaml
-CrewAIAgent:
-  type: object
-  required:
-    - role
-    - goal
-  properties:
-    role:
-      type: string
-      description: "Agent's role/persona"
-      examples:
-        - "Senior Research Analyst"
-        - "Technical Writer"
-        - "Quality Assurance Engineer"
-    goal:
-      type: string
-      description: "Agent's primary objective"
-    backstory:
-      type: string
-      description: "Background context for the agent persona"
-    llm:
-      type: object
-      description: "LLM configuration for this agent"
-      properties:
-        provider:
-          type: string
-          enum: [openai, anthropic, google, azure, ollama, groq]
-        model:
-          type: string
-        temperature:
-          type: number
-          minimum: 0
-          maximum: 2
-    tools:
-      type: array
-      description: "Tools available to this agent"
-      items:
-        type: string
-    allow_delegation:
-      type: boolean
-      default: true
-      description: "Can this agent delegate to others"
-    max_iter:
-      type: integer
-      default: 25
-      minimum: 1
-      description: "Maximum reasoning iterations"
-    max_rpm:
-      type: integer
-      description: "Rate limit for this agent"
-    max_execution_time:
-      type: integer
-      description: "Maximum execution time in seconds"
-    verbose:
-      type: boolean
-      default: false
-      description: "Enable verbose logging"
-    cache:
-      type: boolean
-      default: true
-      description: "Enable response caching"
-    system_template:
-      type: string
-      description: "Custom system prompt template"
-    prompt_template:
-      type: string
-      description: "Custom prompt template"
-    response_template:
-      type: string
-      description: "Custom response template"
-    allow_code_execution:
-      type: boolean
-      default: false
-      description: "Allow code execution capability"
-    code_execution_mode:
-      type: string
-      enum: [safe, unsafe]
-      default: safe
-      description: "Code execution security mode"
-    ossa_agent_ref:
-      type: string
-      description: "Reference to OSSA Agent manifest file"
-      examples:
-        - "./agents/researcher.ossa.yaml"
-        - "ossa://agents/content-writer"
-```
-### CrewAITask
-```yaml
-CrewAITask:
-  type: object
-  required:
-    - description
-    - agent
-  properties:
-    description:
-      type: string
-      description: "Detailed task description"
-    agent:
-      type: string
-      description: "Agent role assigned to this task"
-    expected_output:
-      type: string
-      description: "Description of expected output format"
-    tools:
-      type: array
-      items:
-        type: string
-      description: "Specific tools for this task"
-    context:
-      type: array
-      items:
-        type: string
-      description: "References to other tasks providing context"
-    async_execution:
-      type: boolean
-      default: false
-      description: "Execute asynchronously"
-    output_json:
-      type: object
-      description: "JSON schema for structured output"
-    output_pydantic:
-      type: string
-      description: "Pydantic model class name for validation"
-    output_file:
-      type: string
-      description: "File path for output"
-    human_input:
-      type: boolean
-      default: false
-      description: "Require human input before completion"
-    converter_cls:
-      type: string
-      description: "Custom output converter class"
-    ossa_task_ref:
-      type: string
-      description: "Reference to OSSA Task manifest file"
-      examples:
-        - "./tasks/research.ossa.yaml"
-```
-### CrewAIMemory
-```yaml
-CrewAIMemory:
-  type: object
-  description: "Memory system configuration"
-  properties:
-    enabled:
-      type: boolean
-      default: false
-      description: "Enable memory system"
-    short_term:
-      type: object
-      description: "Short-term memory (within session)"
-      properties:
-        provider:
-          type: string
-          enum: [rag, simple, custom]
-          default: rag
-        embedder:
-          type: object
-          properties:
-            provider:
-              type: string
-              enum: [openai, cohere, google, huggingface]
-            model:
-              type: string
-            config:
-              type: object
-              additionalProperties: true
-    long_term:
-      type: object
-      description: "Long-term memory (persistent)"
-      properties:
-        provider:
-          type: string
-          enum: [rag, sqlite, custom]
-          default: rag
-        storage:
-          type: object
-          properties:
-            type:
-              type: string
-              enum: [chroma, qdrant, pinecone, weaviate, pgvector]
-            connection:
-              type: string
-              description: "Connection string or path"
-    entity:
-      type: object
-      description: "Entity memory (extracted entities)"
-      properties:
-        enabled:
-          type: boolean
-          default: false
-        provider:
-          type: string
-          enum: [rag, spacy, custom]
-    contextual:
-      type: object
-      description: "Contextual memory (RAG-based)"
-      properties:
-        enabled:
-          type: boolean
-          default: false
-        retriever:
-          type: object
-          properties:
-            k:
-              type: integer
-              default: 5
-              description: "Number of results to retrieve"
-            threshold:
-              type: number
-              default: 0.7
-              description: "Similarity threshold"
-```
-### CrewAICallbacks
-```yaml
-CrewAICallbacks:
-  type: object
-  description: "Callback configuration for observability"
-  properties:
-    on_step:
-      type: object
-      description: "Step completion callback"
-      properties:
-        handler:
-          type: string
-          description: "Handler function/class reference"
-        otel_export:
-          type: boolean
-          default: true
-          description: "Export to OpenTelemetry"
-        include_output:
-          type: boolean
-          default: false
-          description: "Include step output in trace"
-    on_task:
-      type: object
-      description: "Task completion callback"
-      properties:
-        handler:
-          type: string
-        otel_export:
-          type: boolean
-          default: true
-        include_output:
-          type: boolean
-          default: false
-    on_crew_start:
-      type: object
-      description: "Crew start callback"
-      properties:
-        handler:
-          type: string
-        otel_export:
-          type: boolean
-          default: true
-    on_crew_end:
-      type: object
-      description: "Crew completion callback"
-      properties:
-        handler:
-          type: string
-        otel_export:
-          type: boolean
-          default: true
-        include_final_output:
-          type: boolean
-          default: true
-    on_agent_action:
-      type: object
-      description: "Agent action callback"
-      properties:
-        handler:
-          type: string
-        otel_export:
-          type: boolean
-          default: true
-    on_tool_use:
-      type: object
-      description: "Tool usage callback"
-      properties:
-        handler:
-          type: string
-        otel_export:
-          type: boolean
-          default: true
-        track_latency:
-          type: boolean
-          default: true
-```
-## Example Manifests
-### Example 1: Research Crew as OSSA Workflow
-CrewAI Python code:
-```python
-from crewai import Agent, Task, Crew, Process
-researcher = Agent(
-    role="Senior Research Analyst",
-    goal="Uncover cutting-edge developments in AI",
-    backstory="You are a seasoned researcher at a tech think tank.",
-    tools=[SerperDevTool(), ScrapeWebsiteTool()],
-    allow_delegation=False,
-    verbose=True
-)
-writer = Agent(
-    role="Tech Content Writer",
-    goal="Write engaging tech content based on research",
-    backstory="You are a renowned content writer for tech blogs.",
-    tools=[],
-    verbose=True
-)
-research_task = Task(
-    description="Research the latest AI developments in 2024",
-    agent=researcher,
-    expected_output="A comprehensive list of AI developments"
-)
-write_task = Task(
-    description="Write a blog post about the AI developments",
-    agent=writer,
-    expected_output="A blog post in markdown format",
-    context=[research_task]
-)
-crew = Crew(
-    agents=[researcher, writer],
-    tasks=[research_task, write_task],
-    process=Process.sequential,
-    memory=True,
-    verbose=True
-)
-```
-Equivalent OSSA Workflow:
-```yaml
-apiVersion: ossa/v0.3.3
-kind: Workflow
-metadata:
-  name: ai-research-crew
-  version: 1.0.0
-  description: "Research and write about AI developments"
-  labels:
-    framework: crewai
-    domain: research
-spec:
-  triggers:
-    - type: manual
-  inputs:
-    type: object
-    properties:
-      topic:
-        type: string
-        default: "AI developments"
-      year:
-        type: integer
-        default: 2024
-  steps:
-    - id: research
-      name: "Research AI Developments"
-      kind: Agent
-      ref: "./agents/researcher.ossa.yaml"
-      input:
-        topic: "${{ workflow.input.topic }}"
-        year: "${{ workflow.input.year }}"
-      output:
-        to: research_results
-    - id: write
-      name: "Write Blog Post"
-      kind: Agent
-      ref: "./agents/writer.ossa.yaml"
-      depends_on:
-        - research
-      input:
-        research: "${{ steps.research.output }}"
-      output:
-        to: blog_post
-  outputs:
-    type: object
-    properties:
-      blog_post:
-        type: string
-  observability:
-    tracing:
-      enabled: true
-      propagate_context: true
-    metrics:
-      enabled: true
-      custom_labels:
-        crew: ai-research-crew
-extensions:
-  crewai:
-    crew_name: ai-research-crew
-    process_type: sequential
-    delegation_enabled: false
-    verbose: true
-    memory_config:
-      enabled: true
-      short_term:
-        provider: rag
-      long_term:
-        provider: rag
-    callbacks:
-      on_step:
-        otel_export: true
-      on_task:
-        otel_export: true
-      on_crew_end:
-        include_final_output: true
-    agents:
-      - role: "Senior Research Analyst"
-        goal: "Uncover cutting-edge developments in AI"
-        backstory: "You are a seasoned researcher at a tech think tank."
-        tools:
-          - web_search
-          - web_scrape
-        allow_delegation: false
-        verbose: true
-        ossa_agent_ref: "./agents/researcher.ossa.yaml"
-      - role: "Tech Content Writer"
-        goal: "Write engaging tech content based on research"
-        backstory: "You are a renowned content writer for tech blogs."
-        tools: []
-        verbose: true
-        ossa_agent_ref: "./agents/writer.ossa.yaml"
-    tasks:
-      - description: "Research the latest AI developments in 2024"
-        agent: "Senior Research Analyst"
-        expected_output: "A comprehensive list of AI developments"
-        ossa_task_ref: "./tasks/research.ossa.yaml"
-      - description: "Write a blog post about the AI developments"
-        agent: "Tech Content Writer"
-        expected_output: "A blog post in markdown format"
-        context:
-          - "Research the latest AI developments in 2024"
-        ossa_task_ref: "./tasks/write-blog.ossa.yaml"
-```
-### Example 2: Researcher Agent Manifest
-```yaml
-apiVersion: ossa/v0.3.3
-kind: Agent
-metadata:
-  name: researcher
-  version: 1.0.0
-  description: "Senior Research Analyst - uncovers cutting-edge developments"
-  labels:
-    framework: crewai
-    crew: ai-research-crew
-  annotations:
-    crewai.goal: "Uncover cutting-edge developments in AI"
-spec:
-  role: |
-    You are a Senior Research Analyst at a leading tech think tank.
-    Your primary goal is to uncover cutting-edge developments in AI.
-    You are known for your thoroughness and ability to find key insights.
-  prompts:
-    system:
-      template: |
-        You are a seasoned researcher at a tech think tank.
-        Your analysis should be comprehensive, data-driven, and insightful.
-        Always cite your sources and provide verifiable information.
-  llm:
-    provider: openai
-    model: gpt-4o
-    temperature: 0.7
-  tools:
-    - type: http
-      name: serper-search
-      config:
-        endpoint: "https://google.serper.dev/search"
-        method: POST
-    - type: http
-      name: web-scraper
-      config:
-        endpoint: "internal://scraper"
-  autonomy:
-    level: semi
-    max_iterations: 25
-    checkpoint_frequency: 5
-  constraints:
-    rate_limit:
-      requests_per_minute: 60
-    resource_limits:
-      max_execution_time_seconds: 300
-  state:
-    persistence:
-      enabled: true
-      type: memory
-    context_window: 10
-  observability:
-    logging:
-      level: debug
-    tracing:
-      enabled: true
-    metrics:
-      enabled: true
-  delegation:
-    enabled: false
-  taxonomy:
-    domain: data
-    subdomain: research
-    capability: analyze_trends
-    concerns:
-      - quality
-      - reliability
-  access:
-    tier: tier_1_read
-    permissions:
-      - read_code
-      - read_configs
-      - execute_queries
-    audit_level: detailed
-runtime:
-  type: unified
-  supports:
-    - local-execution
-    - kubernetes
-```
-### Example 3: Hierarchical Crew Pattern
-```yaml
-apiVersion: ossa/v0.3.3
-kind: Workflow
-metadata:
-  name: content-production-crew
-  version: 1.0.0
-  description: "Hierarchical content production with manager delegation"
-  labels:
-    framework: crewai
-    process: hierarchical
-spec:
-  steps:
-    - id: manager-orchestration
-      name: "Manager Orchestration"
-      kind: Parallel
-      parallel:
-        - id: researcher-1
-          kind: Agent
-          ref: "./agents/researcher.ossa.yaml"
-          input:
-            topic: "Technical trends"
-        - id: researcher-2
-          kind: Agent
-          ref: "./agents/researcher.ossa.yaml"
-          input:
-            topic: "Market analysis"
-        - id: analyst
-          kind: Agent
-          ref: "./agents/analyst.ossa.yaml"
-          input:
-            topic: "Competitive landscape"
-    - id: synthesis
-      kind: Agent
-      ref: "./agents/synthesizer.ossa.yaml"
-      depends_on:
-        - manager-orchestration
-      input:
-        research_1: "${{ steps.researcher-1.output }}"
-        research_2: "${{ steps.researcher-2.output }}"
-        analysis: "${{ steps.analyst.output }}"
-    - id: review
-      kind: Agent
-      ref: "./agents/reviewer.ossa.yaml"
-      depends_on:
-        - synthesis
-      input:
-        content: "${{ steps.synthesis.output }}"
-extensions:
-  crewai:
-    crew_name: content-production-crew
-    process_type: hierarchical
-    delegation_enabled: true
-    manager_llm:
-      provider: anthropic
-      model: claude-sonnet-4-20250514
-      temperature: 0.3
-    memory_config:
-      enabled: true
-      short_term:
-        provider: rag
-      long_term:
-        provider: rag
-        storage:
-          type: chroma
-          connection: "./data/chroma_db"
-    callbacks:
-      on_step:
-        handler: "metrics.CrewStepCallback"
-        otel_export: true
-      on_tool_use:
-        track_latency: true
-```
-### Example 4: OSSA to CrewAI Conversion
-Given an OSSA Workflow, convert to CrewAI Python code:
-```python
-# Auto-generated from OSSA manifest: ai-research-crew.ossa.yaml
-from crewai import Agent, Task, Crew, Process
-from crewai.tools import SerperDevTool, ScrapeWebsiteTool
-# Agent: researcher (from ./agents/researcher.ossa.yaml)
-researcher = Agent(
-    role="Senior Research Analyst",
-    goal="Uncover cutting-edge developments in AI",
-    backstory="""You are a seasoned researcher at a tech think tank.
-Your analysis should be comprehensive, data-driven, and insightful.
-Always cite your sources and provide verifiable information.""",
-    llm="gpt-4o",
-    tools=[SerperDevTool(), ScrapeWebsiteTool()],
-    allow_delegation=False,
-    verbose=True,
-    max_iter=25,
-    cache=True
-)
-# Agent: writer (from ./agents/writer.ossa.yaml)
-writer = Agent(
-    role="Tech Content Writer",
-    goal="Write engaging tech content based on research",
-    backstory="You are a renowned content writer for tech blogs.",
-    llm="gpt-4o",
-    tools=[],
-    verbose=True,
-    max_iter=25
-)
-# Task: research (from ./tasks/research.ossa.yaml)
-research_task = Task(
-    description="Research the latest AI developments in 2024",
-    agent=researcher,
-    expected_output="A comprehensive list of AI developments"
-)
-# Task: write-blog (from ./tasks/write-blog.ossa.yaml)
-write_task = Task(
-    description="Write a blog post about the AI developments",
-    agent=writer,
-    expected_output="A blog post in markdown format",
-    context=[research_task]
-)
-# Crew: ai-research-crew
-crew = Crew(
-    agents=[researcher, writer],
-    tasks=[research_task, write_task],
-    process=Process.sequential,
-    memory=True,
-    verbose=True,
-    max_rpm=60
-)
-# Execute
-result = crew.kickoff()
-```
-## Process Type Mappings
-### Sequential Process
-```yaml
-# CrewAI
-process: Process.sequential
-# OSSA Workflow
-spec:
-  steps:
-    - id: step-1
-      # ...
-    - id: step-2
-      depends_on:
-        - step-1
-    - id: step-3
-      depends_on:
-        - step-2
-```
-### Hierarchical Process
-```yaml
-# CrewAI
-process: Process.hierarchical
-manager_llm: ChatOpenAI(model="gpt-4")
-# OSSA Workflow
-spec:
-  steps:
-    - id: manager
-      kind: Agent
-      ref: "./agents/manager.ossa.yaml"
-      input:
-        available_agents:
-          - researcher
-          - writer
-          - analyst
-    - id: workers
-      kind: Parallel
-      depends_on:
-        - manager
-      parallel:
-        - id: researcher
-          kind: Agent
-          ref: "./agents/researcher.ossa.yaml"
-          condition: "${{ steps.manager.output.delegate_to contains 'researcher' }}"
-        - id: writer
-          kind: Agent
-          ref: "./agents/writer.ossa.yaml"
-          condition: "${{ steps.manager.output.delegate_to contains 'writer' }}"
-extensions:
-  crewai:
-    process_type: hierarchical
-    manager_llm:
-      provider: openai
-      model: gpt-4
-```
-### Consensual Process (Future)
-```yaml
-# OSSA Workflow pattern for consensus
-spec:
-  steps:
-    - id: parallel-analysis
-      kind: Parallel
-      parallel:
-        - id: agent-1
-          kind: Agent
-          ref: "./agents/analyst-1.ossa.yaml"
-        - id: agent-2
-          kind: Agent
-          ref: "./agents/analyst-2.ossa.yaml"
-        - id: agent-3
-          kind: Agent
-          ref: "./agents/analyst-3.ossa.yaml"
-    - id: vote
-      kind: Task
-      ref: "./tasks/consensus-vote.ossa.yaml"
-      depends_on:
-        - parallel-analysis
-      input:
-        votes:
-          - "${{ steps.agent-1.output.decision }}"
-          - "${{ steps.agent-2.output.decision }}"
-          - "${{ steps.agent-3.output.decision }}"
-    - id: final-decision
-      kind: Conditional
-      depends_on:
-        - vote
-      branches:
-        - condition: "${{ steps.vote.output.consensus == true }}"
-          steps:
-            - id: execute-decision
-              kind: Task
-              ref: "./tasks/execute.ossa.yaml"
-      else:
-        - id: escalate
-          kind: Task
-          ref: "./tasks/escalate.ossa.yaml"
-```
-## Delegation Configuration
-### CrewAI Delegation
-```python
-# CrewAI allows delegation between agents
-agent = Agent(
-    role="Manager",
-    allow_delegation=True,
-    # When True, agent can delegate tasks to other agents
-)
-```
-### OSSA Delegation Mapping
-```yaml
-spec:
-  delegation:
-    enabled: true
-    allowed_tiers:
-      - tier_1_read
-      - tier_2_write_limited
-    allowed_operations:
-      - research
-      - analysis
-      - writing
-    requires:
-      - delegation_token
-      - audit_trail
-      - task_specification
-  separation:
-    role: orchestrator
-    can_delegate_to:
-      - analyzer
-      - worker
-    conflicts_with:
-      - executor
-      - approver
-```
-## Memory System Integration
-### Short-term Memory
-```yaml
-extensions:
-  crewai:
-    memory_config:
-      short_term:
-        provider: rag
-        embedder:
-          provider: openai
-          model: text-embedding-3-small
-# Maps to OSSA
-spec:
-  state:
-    context_window: 10
-    memory:
-      type: conversation
-      max_tokens: 4096
-```
-### Long-term Memory
-```yaml
-extensions:
-  crewai:
-    memory_config:
-      long_term:
-        provider: rag
-        storage:
-          type: chroma
-          connection: "./data/chroma"
-# Maps to OSSA
-spec:
-  state:
-    persistence:
-      enabled: true
-      type: database
-      provider: vector
-      config:
-        store: chroma
-        path: "./data/chroma"
-```
-## Callback Integration with OpenTelemetry
-```yaml
-extensions:
-  crewai:
-    callbacks:
-      on_step:
-        handler: "observability.StepCallback"
-        otel_export: true
-        include_output: false
-      on_task:
-        handler: "observability.TaskCallback"
-        otel_export: true
-      on_crew_end:
-        handler: "observability.CrewCallback"
-        include_final_output: true
-# Maps to OSSA observability
-spec:
-  observability:
-    tracing:
-      enabled: true
-      sample_rate: 1.0
-      exporter: otlp
-      endpoint: "http://localhost:4317"
-    metrics:
-      enabled: true
-      custom_labels:
-        framework: crewai
-        crew: "${{ metadata.name }}"
-    logging:
-      level: info
-      structured: true
-```
-## Zod Validation Schema
-```typescript
-import { z } from 'zod';
-// CrewAI Agent validation
-const CrewAIAgentSchema = z.object({
-  role: z.string().min(1),
-  goal: z.string().min(1),
-  backstory: z.string().optional(),
-  llm: z.object({
-    provider: z.enum(['openai', 'anthropic', 'google', 'azure', 'ollama', 'groq']),
-    model: z.string(),
-    temperature: z.number().min(0).max(2).optional()
-  }).optional(),
-  tools: z.array(z.string()).optional(),
-  allow_delegation: z.boolean().default(true),
-  max_iter: z.number().int().positive().default(25),
-  max_rpm: z.number().int().positive().optional(),
-  verbose: z.boolean().default(false),
-  cache: z.boolean().default(true),
-  ossa_agent_ref: z.string().optional()
-});
-// CrewAI Task validation
-const CrewAITaskSchema = z.object({
-  description: z.string().min(1),
-  agent: z.string().min(1),
-  expected_output: z.string().optional(),
-  tools: z.array(z.string()).optional(),
-  context: z.array(z.string()).optional(),
-  async_execution: z.boolean().default(false),
-  output_json: z.record(z.unknown()).optional(),
-  output_file: z.string().optional(),
-  human_input: z.boolean().default(false),
-  ossa_task_ref: z.string().optional()
-});
-// CrewAI Memory validation
-const CrewAIMemorySchema = z.object({
-  enabled: z.boolean().default(false),
-  short_term: z.object({
-    provider: z.enum(['rag', 'simple', 'custom']).default('rag'),
-    embedder: z.object({
-      provider: z.enum(['openai', 'cohere', 'google', 'huggingface']),
-      model: z.string(),
-      config: z.record(z.unknown()).optional()
-    }).optional()
-  }).optional(),
-  long_term: z.object({
-    provider: z.enum(['rag', 'sqlite', 'custom']).default('rag'),
-    storage: z.object({
-      type: z.enum(['chroma', 'qdrant', 'pinecone', 'weaviate', 'pgvector']),
-      connection: z.string()
-    }).optional()
-  }).optional(),
-  entity: z.object({
-    enabled: z.boolean().default(false),
-    provider: z.enum(['rag', 'spacy', 'custom']).optional()
-  }).optional()
-});
-// CrewAI Callbacks validation
-const CrewAICallbacksSchema = z.object({
-  on_step: z.object({
-    handler: z.string().optional(),
-    otel_export: z.boolean().default(true),
-    include_output: z.boolean().default(false)
-  }).optional(),
-  on_task: z.object({
-    handler: z.string().optional(),
-    otel_export: z.boolean().default(true),
-    include_output: z.boolean().default(false)
-  }).optional(),
-  on_crew_start: z.object({
-    handler: z.string().optional(),
-    otel_export: z.boolean().default(true)
-  }).optional(),
-  on_crew_end: z.object({
-    handler: z.string().optional(),
-    otel_export: z.boolean().default(true),
-    include_final_output: z.boolean().default(true)
-  }).optional()
-});
-// Full CrewAI Extension validation
-export const CrewAIExtensionSchema = z.object({
-  crew_name: z.string().regex(/^[a-z][a-z0-9_-]*$/),
-  agents: z.array(CrewAIAgentSchema).optional(),
-  tasks: z.array(CrewAITaskSchema).optional(),
-  process_type: z.enum(['sequential', 'hierarchical', 'consensual']).default('sequential'),
-  delegation_enabled: z.boolean().default(true),
-  memory_config: CrewAIMemorySchema.optional(),
-  manager_llm: z.object({
-    provider: z.enum(['openai', 'anthropic', 'google', 'azure', 'ollama']),
-    model: z.string(),
-    temperature: z.number().min(0).max(2).optional()
-  }).optional(),
-  max_rpm: z.number().int().positive().optional(),
-  share_crew: z.boolean().default(false),
-  callbacks: CrewAICallbacksSchema.optional(),
-  verbose: z.boolean().default(false),
-  language: z.string().default('en')
-});
-// Type inference
-export type CrewAIExtension = z.infer<typeof CrewAIExtensionSchema>;
-export type CrewAIAgent = z.infer<typeof CrewAIAgentSchema>;
-export type CrewAITask = z.infer<typeof CrewAITaskSchema>;
-```
-## OpenAPI Specification
-```yaml
-openapi: 3.1.0
-info:
-  title: OSSA CrewAI Extension API
-  version: 0.3.3
-paths:
-  /api/v1/crewai/crews:
-    get:
-      summary: List all CrewAI crews
-      operationId: listCrews
-      responses:
-        '200':
-          description: List of crews
-          content:
-            application/json:
-              schema:
-                type: array
-                items:
-                  $ref: '#/components/schemas/CrewSummary'
-    post:
-      summary: Create a new crew from OSSA workflow
-      operationId: createCrew
-      requestBody:
-        required: true
-        content:
-          application/json:
-            schema:
-              $ref: '#/components/schemas/CreateCrewRequest'
-      responses:
-        '201':
-          description: Crew created
-          content:
-            application/json:
-              schema:
-                $ref: '#/components/schemas/Crew'
-  /api/v1/crewai/crews/{crewId}:
-    get:
-      summary: Get crew details
-      operationId: getCrew
-      parameters:
-        - name: crewId
-          in: path
-          required: true
-          schema:
-            type: string
-      responses:
-        '200':
-          description: Crew details
-          content:
-            application/json:
-              schema:
-                $ref: '#/components/schemas/Crew'
-    put:
-      summary: Update crew
-      operationId: updateCrew
-      parameters:
-        - name: crewId
-          in: path
-          required: true
-          schema:
-            type: string
-      requestBody:
-        required: true
-        content:
-          application/json:
-            schema:
-              $ref: '#/components/schemas/UpdateCrewRequest'
-      responses:
-        '200':
-          description: Crew updated
-    delete:
-      summary: Delete crew
-      operationId: deleteCrew
-      parameters:
-        - name: crewId
-          in: path
-          required: true
-          schema:
-            type: string
-      responses:
-        '204':
-          description: Crew deleted
-  /api/v1/crewai/crews/{crewId}/kickoff:
-    post:
-      summary: Execute crew
-      operationId: kickoffCrew
-      parameters:
-        - name: crewId
-          in: path
-          required: true
-          schema:
-            type: string
-      requestBody:
-        content:
-          application/json:
-            schema:
-              type: object
-              properties:
-                inputs:
-                  type: object
-                  additionalProperties: true
-      responses:
-        '202':
-          description: Crew execution started
-          content:
-            application/json:
-              schema:
-                $ref: '#/components/schemas/CrewExecution'
-  /api/v1/crewai/convert/to-ossa:
-    post:
-      summary: Convert CrewAI Python to OSSA manifest
-      operationId: convertToOSSA
-      requestBody:
-        required: true
-        content:
-          application/json:
-            schema:
-              type: object
-              properties:
-                python_code:
-                  type: string
-                  description: CrewAI Python code
-      responses:
-        '200':
-          description: OSSA manifest
-          content:
-            application/yaml:
-              schema:
-                type: string
-  /api/v1/crewai/convert/from-ossa:
-    post:
-      summary: Convert OSSA manifest to CrewAI Python
-      operationId: convertFromOSSA
-      requestBody:
-        required: true
-        content:
-          application/yaml:
-            schema:
-              type: string
-      responses:
-        '200':
-          description: CrewAI Python code
-          content:
-            text/x-python:
-              schema:
-                type: string
-components:
-  schemas:
-    CrewSummary:
-      type: object
-      properties:
-        id:
-          type: string
-        name:
-          type: string
-        process_type:
-          type: string
-          enum: [sequential, hierarchical, consensual]
-        agent_count:
-          type: integer
-        task_count:
-          type: integer
-        created_at:
-          type: string
-          format: date-time
-    Crew:
-      type: object
-      properties:
-        id:
-          type: string
-        manifest:
-          $ref: '#/components/schemas/OSSAWorkflow'
-        crewai_extension:
-          $ref: '#/components/schemas/CrewAIExtension'
-    CrewAIExtension:
-      type: object
-      properties:
-        crew_name:
-          type: string
-        agents:
-          type: array
-          items:
-            $ref: '#/components/schemas/CrewAIAgent'
-        tasks:
-          type: array
-          items:
-            $ref: '#/components/schemas/CrewAITask'
-        process_type:
-          type: string
-          enum: [sequential, hierarchical, consensual]
-        delegation_enabled:
-          type: boolean
-        memory_config:
-          $ref: '#/components/schemas/CrewAIMemory'
-    CrewAIAgent:
-      type: object
-      required:
-        - role
-        - goal
-      properties:
-        role:
-          type: string
-        goal:
-          type: string
-        backstory:
-          type: string
-        tools:
-          type: array
-          items:
-            type: string
-        allow_delegation:
-          type: boolean
-        max_iter:
-          type: integer
-    CrewAITask:
-      type: object
-      required:
-        - description
-        - agent
-      properties:
-        description:
-          type: string
-        agent:
-          type: string
-        expected_output:
-          type: string
-        context:
-          type: array
-          items:
-            type: string
-    CrewAIMemory:
-      type: object
-      properties:
-        enabled:
-          type: boolean
-        short_term:
-          type: object
-        long_term:
-          type: object
-    CrewExecution:
-      type: object
-      properties:
-        execution_id:
-          type: string
-        status:
-          type: string
-          enum: [pending, running, completed, failed]
-        started_at:
-          type: string
-          format: date-time
-        trace_id:
-          type: string
-          description: OpenTelemetry trace ID
-    OSSAWorkflow:
-      type: object
-      description: Full OSSA Workflow manifest
-```
-## Related
-- [CrewAI Documentation](https://docs.crewai.com/)
-- [OSSA v0.3.3 Specification](../ossa-0.3.3.schema.json)
-- [OSSA Workflow Kind](../UNIFIED-SCHEMA.md)
-- [Agent Taxonomy](../taxonomy.yaml)
-- [Delegation Configuration](../access_tiers.yaml)