0pt/src/document-workbench/candidate-catalog.ts

/**
 * Product planning inventory, NOT an execution registry. Every entry needs
 * version pinning, licence/model review, installation and integration tests
 * before a separately authorized DocumentTool may be created for it.
 */
export type CandidateRole =
  | 'detection'
  | 'document'
  | 'ocr'
  | 'table'
  | 'quality'
  | 'mapping'
  | 'entity'
  | 'media'
  | 'chat'
  | 'workflow'
  | 'metering'
  | 'model'
  | 'integration'
  | 'agent'
  | 'decision'
  | 'evaluation'
  | 'observability'
  | 'browser';
export interface ToolCandidate {
  id: string;
  name: string;
  source: string;
  roles: readonly CandidateRole[];
  /** Planning only; no candidate is automatically allowed to execute. */
  status: 'qualification_required';
  review: 'standard' | 'additional_license_or_service_review';
}

const candidate = (
  id: string,
  name: string,
  source: string,
  roles: readonly CandidateRole[],
  review: ToolCandidate['review'] = 'standard',
): ToolCandidate => ({ id, name, source, roles, status: 'qualification_required', review });

export const TOOL_CANDIDATES: readonly ToolCandidate[] = [
  candidate('tika', 'Apache Tika', 'https://github.com/apache/tika', ['detection', 'document']),
  candidate('docling', 'Docling', 'https://github.com/docling-project/docling', [
    'detection',
    'document',
    'table',
  ]),
  candidate('unstructured', 'Unstructured', 'https://github.com/Unstructured-IO/unstructured', [
    'document',
    'table',
  ]),
  candidate('pdfplumber', 'pdfplumber', 'https://github.com/jsvine/pdfplumber', ['document', 'table']),
  candidate('tesseract', 'Tesseract', 'https://github.com/tesseract-ocr/tesseract', ['ocr']),
  candidate('paddleocr', 'PaddleOCR', 'https://github.com/PaddlePaddle/PaddleOCR', ['ocr', 'document']),
  candidate('ocrmypdf', 'OCRmyPDF', 'https://github.com/ocrmypdf/OCRmyPDF', ['ocr', 'document']),
  candidate('markitdown', 'MarkItDown', 'https://github.com/microsoft/markitdown', ['document']),
  candidate('duckdb', 'DuckDB', 'https://github.com/duckdb/duckdb', ['table', 'quality']),
  candidate('polars', 'Polars', 'https://github.com/pola-rs/polars', ['table', 'quality']),
  candidate('arrow', 'Apache Arrow and Parquet', 'https://github.com/apache/arrow', ['table']),
  candidate('frictionless', 'Frictionless', 'https://github.com/frictionlessdata/frictionless-py', [
    'table',
    'quality',
  ]),
  candidate('pandera', 'Pandera', 'https://github.com/unionai-oss/pandera', ['table', 'quality']),
  candidate(
    'great-expectations',
    'Great Expectations',
    'https://github.com/great-expectations/great_expectations',
    ['quality'],
  ),
  candidate('openrefine', 'OpenRefine', 'https://github.com/OpenRefine/OpenRefine', [
    'table',
    'quality',
    'mapping',
  ]),
  candidate('valentine', 'Valentine', 'https://github.com/delftdata/valentine', ['mapping']),
  candidate('rapidfuzz', 'RapidFuzz', 'https://github.com/rapidfuzz/RapidFuzz', ['entity', 'mapping']),
  candidate('splink', 'Splink', 'https://github.com/moj-analytical-services/splink', ['entity']),
  candidate('sqlglot', 'SQLGlot', 'https://github.com/tobymao/sqlglot', ['table', 'mapping']),
  candidate('assistant-ui', 'assistant-ui', 'https://github.com/assistant-ui/assistant-ui', ['chat']),
  candidate('temporal', 'Temporal', 'https://github.com/temporalio/temporal', ['workflow']),
  candidate('langgraph', 'LangGraph', 'https://github.com/langchain-ai/langgraph', ['workflow', 'agent']),
  candidate(
    'jev',
    'TypeSafe Jev',
    'https://docs.typesafe.ai/',
    ['decision', 'model'],
    'additional_license_or_service_review',
  ),
  candidate('pydantic-ai', 'PydanticAI', 'https://github.com/pydantic/pydantic-ai', ['agent']),
  candidate(
    'microsoft-agent-framework',
    'Microsoft Agent Framework',
    'https://github.com/microsoft/agent-framework',
    ['agent', 'workflow'],
  ),
  candidate('semantic-kernel', 'Semantic Kernel', 'https://github.com/microsoft/semantic-kernel', ['agent']),
  candidate(
    'autogen',
    'AutoGen',
    'https://github.com/microsoft/autogen',
    ['agent'],
    'additional_license_or_service_review',
  ),
  candidate('crewai', 'CrewAI', 'https://github.com/crewAIInc/crewAI', ['agent', 'workflow']),
  candidate('haystack', 'Haystack', 'https://github.com/deepset-ai/haystack', ['agent', 'workflow']),
  candidate(
    'llama-index',
    'LlamaIndex',
    'https://github.com/run-llama/llama_index',
    ['agent', 'document'],
    'additional_license_or_service_review',
  ),
  candidate('llama-agents', 'llama-agents', 'https://github.com/run-llama/llama-agents', [
    'agent',
    'workflow',
  ]),
  candidate('google-adk', 'Google ADK', 'https://github.com/google/adk-python', ['agent', 'workflow']),
  candidate(
    'mastra',
    'Mastra',
    'https://github.com/mastra-ai/mastra',
    ['agent', 'workflow'],
    'additional_license_or_service_review',
  ),
  candidate('openai-agents-sdk', 'OpenAI Agents SDK', 'https://github.com/openai/openai-agents-python', [
    'agent',
  ]),
  candidate('smolagents', 'smolagents', 'https://github.com/huggingface/smolagents', ['agent']),
  candidate('dspy', 'DSPy', 'https://github.com/stanfordnlp/dspy', ['agent', 'evaluation']),
  candidate(
    'flowise',
    'Flowise',
    'https://github.com/FlowiseAI/Flowise',
    ['agent', 'workflow'],
    'additional_license_or_service_review',
  ),
  candidate(
    'browser-use',
    'browser-use',
    'https://github.com/browser-use/browser-use',
    ['browser', 'agent'],
    'additional_license_or_service_review',
  ),
  candidate('playwright-mcp', 'Playwright MCP', 'https://github.com/microsoft/playwright-mcp', ['browser']),
  candidate('promptfoo', 'Promptfoo', 'https://github.com/promptfoo/promptfoo', ['evaluation']),
  candidate('deepeval', 'DeepEval', 'https://github.com/confident-ai/deepeval', ['evaluation']),
  candidate('ragas', 'Ragas', 'https://github.com/vibrantlabsai/ragas', ['evaluation']),
  candidate(
    'langfuse',
    'Langfuse',
    'https://github.com/langfuse/langfuse',
    ['evaluation', 'observability'],
    'additional_license_or_service_review',
  ),
  candidate(
    'arize-phoenix',
    'Arize Phoenix',
    'https://github.com/Arize-ai/phoenix',
    ['evaluation', 'observability'],
    'additional_license_or_service_review',
  ),
  candidate('openmeter', 'OpenMeter', 'https://github.com/openmeterio/openmeter', ['metering']),
  candidate('whisper', 'Whisper', 'https://github.com/openai/whisper', ['media']),
  candidate('faster-whisper', 'faster-whisper', 'https://github.com/SYSTRAN/faster-whisper', ['media']),
  candidate('ffmpeg', 'FFmpeg', 'https://ffmpeg.org/', ['media'], 'additional_license_or_service_review'),
  candidate(
    'pymupdf',
    'PyMuPDF',
    'https://github.com/pymupdf/PyMuPDF',
    ['document'],
    'additional_license_or_service_review',
  ),
  candidate(
    'marker',
    'Marker',
    'https://github.com/datalab-to/marker',
    ['document', 'ocr'],
    'additional_license_or_service_review',
  ),
  candidate(
    'mineru',
    'MinerU',
    'https://github.com/opendatalab/MinerU',
    ['document', 'ocr'],
    'additional_license_or_service_review',
  ),
  candidate(
    'airbyte',
    'Airbyte',
    'https://github.com/airbytehq/airbyte',
    ['integration'],
    'additional_license_or_service_review',
  ),
  candidate(
    'dify',
    'Dify',
    'https://github.com/langgenius/dify',
    ['workflow', 'chat'],
    'additional_license_or_service_review',
  ),
  candidate(
    'n8n',
    'n8n',
    'https://github.com/n8n-io/n8n',
    ['workflow', 'integration'],
    'additional_license_or_service_review',
  ),
  candidate(
    'open-webui',
    'Open WebUI',
    'https://github.com/open-webui/open-webui',
    ['chat'],
    'additional_license_or_service_review',
  ),
  candidate(
    'openrouter',
    'OpenRouter',
    'https://openrouter.ai/',
    ['model'],
    'additional_license_or_service_review',
  ),
];

export function candidatesForRole(role: CandidateRole): readonly ToolCandidate[] {
  return TOOL_CANDIDATES.filter((entry) => entry.roles.includes(role));
}