@datalayer/agent-runtimes 1.0.19 → 1.0.21

package/scripts/codegen/generate_models.py

@@ -104,6 +104,11 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
         else:
             env_vars_formatted = "[]"
 
+        tokens_limit = spec.get("tokens_limit")
+        tokens_limit_formatted = (
+            str(tokens_limit) if tokens_limit is not None else "None"
+        )
+
         lines.extend(
             [
                 f"{const_name} = AIModel(",
@@ -114,6 +119,7 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
                 f'    provider="{spec["provider"]}",',
                 f"    default={spec.get('default', False)},",
                 f"    required_env_vars={env_vars_formatted},",
+                f"    tokens_limit={tokens_limit_formatted},",
                 ")",
                 "",
             ]
@@ -288,20 +294,27 @@ def generate_typescript_code(specs: list[dict[str, Any]]) -> str:
         # Escape description for TypeScript
         description = spec.get("description", "").replace("'", "\\'")
 
-        lines.extend(
+        tokens_limit = spec.get("tokens_limit")
+
+        model_lines = [
+            f"export const {const_name}: AIModel = {{",
+            f"  id: '{spec['id']}',",
+            f"  version: '{spec['version']}',",
+            f"  name: '{spec['name']}',",
+            f"  description: '{description}',",
+            f"  provider: '{spec['provider']}',",
+            f"  default: {str(spec.get('default', False)).lower()},",
+            f"  requiredEnvVars: {env_vars_formatted},",
+        ]
+        if tokens_limit is not None:
+            model_lines.append(f"  tokensLimit: {tokens_limit},")
+        model_lines.extend(
             [
-                f"export const {const_name}: AIModel = {{",
-                f"  id: '{spec['id']}',",
-                f"  version: '{spec['version']}',",
-                f"  name: '{spec['name']}',",
-                f"  description: '{description}',",
-                f"  provider: '{spec['provider']}',",
-                f"  default: {str(spec.get('default', False)).lower()},",
-                f"  requiredEnvVars: {env_vars_formatted},",
                 "};",
                 "",
             ]
         )
+        lines.extend(model_lines)
 
     # Generate catalog object
     lines.extend(

package/scripts/gen_workers.py

@@ -0,0 +1,340 @@
+# Copyright (c) 2025-2026 Datalayer, Inc.
+# Distributed under the terms of the Modified BSD License.
+"""One-off generator for the workers-* agent specs (9 verticals).
+
+Creates one YAML spec per worker example from the Agent Workers portfolio.
+Skips the three flagship workers that already exist as hand-authored specs.
+"""
+
+from pathlib import Path
+
+AGENTS_DIR = Path(__file__).resolve().parent.parent / "agentspecs" / "agentspecs" / "agents"
+
+# Per-vertical defaults: label, emoji, icon, color, mcp servers, skills, envvars.
+VERTICALS = {
+    "earth-observation": {
+        "emoji": "🛰️", "icon": "globe", "color": "#10B981",
+        "mcp": ["earthdata:0.0.1", "tavily:0.0.1"],
+        "skills": ["text-summarizer:0.0.1", "crawl:0.0.1"],
+        "envvars": ["TAVILY_API_KEY:0.0.1"],
+    },
+    "accounting": {
+        "emoji": "🧾", "icon": "credit-card", "color": "#F59E0B",
+        "mcp": ["odoo:0.0.1"],
+        "skills": ["accounting:0.0.1", "pdf:0.0.1"],
+        "envvars": [],
+    },
+    "energy": {
+        "emoji": "⚡", "icon": "zap", "color": "#EAB308",
+        "mcp": ["tavily:0.0.1", "chart:0.0.1"],
+        "skills": ["text-summarizer:0.0.1"],
+        "envvars": ["TAVILY_API_KEY:0.0.1"],
+    },
+    "insurance": {
+        "emoji": "🛡️", "icon": "shield", "color": "#EF4444",
+        "mcp": ["earthdata:0.0.1", "tavily:0.0.1"],
+        "skills": ["text-summarizer:0.0.1", "pdf:0.0.1"],
+        "envvars": ["TAVILY_API_KEY:0.0.1"],
+    },
+    "life-sciences": {
+        "emoji": "🧬", "icon": "beaker", "color": "#06B6D4",
+        "mcp": ["huggingface:0.0.1", "kaggle:0.0.1"],
+        "skills": ["text-summarizer:0.0.1", "pdf:0.0.1"],
+        "envvars": ["HF_TOKEN:0.0.1", "KAGGLE_TOKEN:0.0.1"],
+    },
+    "capital-markets": {
+        "emoji": "📈", "icon": "graph", "color": "#3B82F6",
+        "mcp": ["alphavantage:0.0.1", "chart:0.0.1"],
+        "skills": ["text-summarizer:0.0.1"],
+        "envvars": ["ALPHAVANTAGE_API_KEY:0.0.1"],
+    },
+    "personal-assistant": {
+        "emoji": "🤖", "icon": "person", "color": "#A855F7",
+        "mcp": ["tavily:0.0.1", "google-workspace:0.0.1"],
+        "skills": ["text-summarizer:0.0.1", "crawl:0.0.1", "events:0.0.1"],
+        "envvars": ["TAVILY_API_KEY:0.0.1"],
+    },
+    "marketing": {
+        "emoji": "📣", "icon": "megaphone", "color": "#8B5CF6",
+        "mcp": ["tavily:0.0.1", "slack:0.0.1"],
+        "skills": ["text-summarizer:0.0.1", "crawl:0.0.1", "events:0.0.1"],
+        "envvars": ["TAVILY_API_KEY:0.0.1", "SLACK_BOT_TOKEN:0.0.1"],
+    },
+    "market-analyst": {
+        "emoji": "🔎", "icon": "search", "color": "#EC4899",
+        "mcp": ["tavily:0.0.1", "salesforce:0.0.1"],
+        "skills": ["text-summarizer:0.0.1", "crawl:0.0.1"],
+        "envvars": ["TAVILY_API_KEY:0.0.1"],
+    },
+}
+
+# worker slug -> (vertical, name, description, goal). Renamed flagships excluded.
+WORKERS = {
+    # Earth Observation
+    "change-detection": ("earth-observation", "Change Detection Worker",
+        "Detects environmental and land-use change between satellite image time periods and produces evidence-backed change maps.",
+        "Compare imagery across time periods, detect meaningful land-use and environmental change, and deliver annotated maps and reports."),
+    "crop-monitoring": ("earth-observation", "Crop Monitoring Worker",
+        "Monitors crop health, growth stages, and field conditions using multi-temporal satellite imagery.",
+        "Track crop vigor and growth over time from satellite imagery and flag fields that need attention."),
+    "disaster-assessment": ("earth-observation", "Disaster Assessment Worker",
+        "Assesses affected areas after natural disasters by comparing pre- and post-event satellite imagery.",
+        "Rapidly estimate affected areas and damage extent after a disaster and produce a response-ready assessment."),
+    "infrastructure-monitoring": ("earth-observation", "Infrastructure Monitoring Worker",
+        "Monitors infrastructure and physical assets from satellite imagery and detects anomalies over time.",
+        "Continuously monitor infrastructure from satellite imagery and surface anomalies with supporting evidence."),
+    "environmental-compliance": ("earth-observation", "Environmental Compliance Worker",
+        "Tracks environmental compliance from Earth observation data and generates recurring, auditable reports.",
+        "Detect potential environmental compliance issues from EO data and produce traceable evidence-backed reports."),
+    # Accounting (Odoo)
+    "ap-invoice": ("accounting", "AP Invoice Worker",
+        "Extracts, codes, and matches supplier invoices to purchase orders and receipts in Odoo.",
+        "Extract and code supplier invoices, match them to POs and receipts, and prepare them for approval."),
+    "bank-reconciliation": ("accounting", "Bank Reconciliation Worker",
+        "Reconciles bank statements against the ledger, matches transactions, and investigates exceptions.",
+        "Reconcile bank feeds against the ledger, auto-match transactions, and flag exceptions for review."),
+    "month-end-close": ("accounting", "Month-End Close Worker",
+        "Prepares close checklists, suggests accruals, and reconciles intercompany balances for month-end close.",
+        "Drive the month-end close checklist, suggest accruals, and surface remaining reconciliation gaps."),
+    "collections": ("accounting", "Collections Worker",
+        "Follows up on overdue receivables and drafts collection communications for approval.",
+        "Prioritize overdue receivables and draft tailored collection follow-ups for human approval."),
+    "expense-audit": ("accounting", "Expense Audit Worker",
+        "Audits employee expenses for duplicates, anomalies, and policy violations.",
+        "Review expenses for duplicates, anomalies, and policy breaches, and summarize findings."),
+    "audit-pack-builder": ("accounting", "Audit-Pack Builder",
+        "Assembles auditable evidence packs from invoices, approvals, and postings across Odoo.",
+        "Assemble a complete, traceable audit-evidence pack for a given period or transaction set."),
+    # Energy and Utilities
+    "renewable-asset-performance": ("energy", "Renewable Asset Performance Worker",
+        "Compares expected versus actual generation, adjusts for weather, and investigates performance anomalies.",
+        "Compare expected vs actual generation, weather-adjust performance, and investigate underperformance."),
+    "grid-forecast": ("energy", "Grid Forecast Worker",
+        "Forecasts electricity demand and prices to support grid operations and planning.",
+        "Produce demand and price forecasts for grid operations with clear assumptions and confidence."),
+    "curtailment-investigator": ("energy", "Curtailment Investigator",
+        "Analyzes curtailment events, quantifies lost generation, and investigates root causes.",
+        "Investigate curtailment events, quantify lost generation, and explain the primary drivers."),
+    "energy-trading-analyst": ("energy", "Energy Trading Analyst",
+        "Analyzes power markets, price signals, and trading opportunities across energy markets.",
+        "Analyze power-market signals and surface evidence-backed trading opportunities and risks."),
+    "predictive-maintenance": ("energy", "Predictive Maintenance Worker",
+        "Detects equipment degradation and anomalies to predict failures and prioritize maintenance.",
+        "Detect degradation and anomalies, predict likely failures, and prioritize maintenance work."),
+    # Insurance
+    "cat-exposure": ("insurance", "CAT Exposure Worker",
+        "Geocodes insurance portfolios and overlays hazard data to analyze catastrophe exposure.",
+        "Geocode portfolios, overlay hazard layers, and quantify catastrophe exposure with lineage."),
+    "event-response": ("insurance", "Event Response Worker",
+        "Assesses live catastrophe events against insured portfolios to estimate exposure and losses.",
+        "Assess a live catastrophe event against the portfolio and publish an auditable exposure estimate."),
+    "portfolio-accumulation": ("insurance", "Portfolio Accumulation Worker",
+        "Analyzes accumulation and concentration risk across an insurance portfolio.",
+        "Analyze accumulation and concentration risk and highlight where limits are approached."),
+    "exposure-data-quality": ("insurance", "Exposure Data Quality Worker",
+        "Cleanses exposure data, validates geocoding, and improves data quality for risk analysis.",
+        "Cleanse and validate exposure data and geocoding, and report remaining data-quality gaps."),
+    "model-comparison": ("insurance", "Model Comparison Worker",
+        "Compares catastrophe models and scenarios and explains differences in loss drivers.",
+        "Compare catastrophe models and scenarios and explain the drivers behind divergent results."),
+    # Life Sciences
+    "rna-seq": ("life-sciences", "RNA-Seq Worker",
+        "Runs RNA-seq quality control and differential expression analysis with reproducible pipelines.",
+        "Run RNA-seq QC and differential expression analysis and produce reproducible, documented results."),
+    "variant-analysis": ("life-sciences", "Variant Analysis Worker",
+        "Analyzes genomic variants, annotates findings, and prepares interpretable reports.",
+        "Analyze and annotate genomic variants and produce an interpretable, provenance-tracked report."),
+    "single-cell-processing": ("life-sciences", "Single-Cell Processing Worker",
+        "Processes, clusters, and annotates single-cell datasets with reproducible workflows.",
+        "Process and cluster single-cell data and deliver annotated, reproducible outputs."),
+    "cohort-comparison": ("life-sciences", "Cohort Comparison Worker",
+        "Compares cohorts, validates statistical results, and checks reproducibility.",
+        "Compare cohorts, validate the statistics, and confirm results are reproducible."),
+    "pipeline-debugger": ("life-sciences", "Pipeline Debugger",
+        "Troubleshoots bioinformatics pipeline failures and suggests corrective actions.",
+        "Diagnose bioinformatics pipeline failures and recommend concrete fixes with evidence."),
+    "compute-cost-optimizer": ("life-sciences", "Compute-Cost Optimizer",
+        "Optimizes pipeline runtime and compute cost while preserving reproducibility.",
+        "Analyze pipeline runs and recommend runtime and cost optimizations without breaking reproducibility."),
+    # Capital Markets
+    "quant-research": ("capital-markets", "Quant Research Worker",
+        "Constructs datasets and researches factors for quantitative strategies with point-in-time care.",
+        "Construct clean datasets and research factors, respecting point-in-time correctness."),
+    "portfolio-risk": ("capital-markets", "Portfolio Risk Worker",
+        "Computes exposures and generates recurring, auditable portfolio risk reports.",
+        "Compute portfolio exposures and produce recurring, auditable risk reports."),
+    "backtest-auditor": ("capital-markets", "Backtest Auditor",
+        "Audits backtests for look-ahead leakage, survivorship bias, and methodological errors.",
+        "Audit a backtest for leakage, survivorship bias, and other flaws, and report findings."),
+    "factor-analysis": ("capital-markets", "Factor Analysis Worker",
+        "Analyzes factor exposures and returns across a portfolio or strategy.",
+        "Analyze factor exposures and returns and explain the primary contributors."),
+    "performance-attribution": ("capital-markets", "Performance Attribution Worker",
+        "Attributes portfolio performance to its underlying drivers and produces clear reports.",
+        "Attribute portfolio performance to its drivers and produce a clear, auditable report."),
+    "scenario-testing": ("capital-markets", "Scenario Testing Worker",
+        "Runs stress tests and scenario analysis to quantify portfolio sensitivity to shocks.",
+        "Run stress tests and scenarios and quantify portfolio sensitivity to defined shocks."),
+    # Personal Assistant (news-aggregator excluded, already exists)
+    "travel-recommender": ("personal-assistant", "Travel Recommender",
+        "Researches travel options and builds personalized itineraries with source comparison.",
+        "Research travel options and build a personalized itinerary, comparing sources and prices."),
+    "trends-seeker": ("personal-assistant", "Trends Seeker",
+        "Tracks emerging trends across sources and summarizes what is gaining momentum.",
+        "Track emerging trends across sources and summarize what is worth attention and why."),
+    "product-finder": ("personal-assistant", "Product Finder",
+        "Researches and compares products against a user's criteria and budget.",
+        "Research and compare products against the user's criteria and recommend the best fit."),
+    "job-hunter": ("personal-assistant", "Job Hunter",
+        "Searches for relevant jobs and prepares tailored application materials.",
+        "Find relevant roles and prepare tailored application materials for review."),
+    "coding-tutor": ("personal-assistant", "Coding Tutor",
+        "Teaches coding through interactive, executable exercises and step-by-step feedback.",
+        "Teach coding interactively with executable exercises and give clear, incremental feedback."),
+    "mail-triage": ("personal-assistant", "Mail Triage Worker",
+        "Classifies the inbox, drafts replies, and extracts follow-up tasks, with approval for sending.",
+        "Classify the inbox, draft replies, and extract tasks, requiring approval before sending anything."),
+    # Marketing (social-marketer excluded, already exists)
+    "campaign-planning": ("marketing", "Campaign Planning Worker",
+        "Plans content calendars and campaign tasks across channels.",
+        "Plan a content calendar and campaign tasks aligned to goals and channels."),
+    "content-repurposing": ("marketing", "Content Repurposing Worker",
+        "Adapts existing content and assets across channels and formats.",
+        "Repurpose existing content into channel-specific formats while keeping brand voice."),
+    "social-listening": ("marketing", "Social Listening Worker",
+        "Monitors conversations across platforms and surfaces actionable insights.",
+        "Monitor conversations at scale and surface actionable, evidence-backed insights."),
+    "community-response": ("marketing", "Community Response Worker",
+        "Triages comments and messages and drafts community replies for approval.",
+        "Triage community comments and draft on-brand replies for human approval."),
+    # Market Analyst (customer-interviewer excluded, already exists)
+    "interview-guide": ("market-analyst", "Interview Guide Worker",
+        "Generates research plans and structured interview guides from research objectives.",
+        "Turn research objectives into a structured research plan and interview guide."),
+    "research-recruiter": ("market-analyst", "Research Recruiter",
+        "Selects and schedules research participants and manages consent.",
+        "Select suitable participants, schedule sessions, and manage consent."),
+    "thematic-analysis": ("market-analyst", "Thematic Analysis Worker",
+        "Codes transcripts, extracts themes, and organizes qualitative evidence.",
+        "Code transcripts, extract themes, and organize the supporting evidence."),
+    "competitive-intelligence": ("market-analyst", "Competitive Intelligence Worker",
+        "Researches competitors and market dynamics and synthesizes findings.",
+        "Research competitors and market dynamics and synthesize evidence-backed findings."),
+    "evidence-repository": ("market-analyst", "Evidence Repository Worker",
+        "Organizes research evidence for traceability from qualitative findings to conclusions.",
+        "Organize research evidence so every conclusion traces back to its source."),
+}
+
+TEMPLATE = """# Copyright (c) 2025-2026 Datalayer, Inc.
+# Distributed under the terms of the Modified BSD License.
+
+# Agent Specification: {name}
+# {vertical_label} vertical worker.
+
+id: workers-{slug}
+version: 0.0.1
+name: {name}
+domain: {vertical}
+description: >-
+  {description}
+
+tags:
+{tags}
+
+enabled: true
+model: "bedrock:us.anthropic.claude-sonnet-4-6"
+
+sandbox_variant: jupyter
+
+# Durable memory to retain context across sessions
+memory: mem0
+
+# MCP servers used by this agent
+mcp_servers:
+{mcp}
+
+# Skills available to this agent
+skills:
+{skills}
+
+# Runtime tools — sensitive actions require human approval
+tools:
+  - runtime-echo:0.0.1
+  - runtime-sensitive-echo:0.0.1
+frontend_tools:
+  - jupyter-notebook:0.0.1
+  - lexical-document:0.0.1
+{envvars}
+# Runtime environment
+environment_name: ai-agents-env
+
+# UI customization
+icon: {icon}
+emoji: "{emoji}"
+color: "{color}"
+
+# Goal / Prompt (Step 4 in the UI — the user-facing objective)
+goal: >-
+  {goal}
+
+# Protocols & UI Extensions
+protocol: vercel-ai
+ui_extension: a2ui
+
+# Model configuration
+model_config:
+  temperature: 0.5
+  max_tokens: 4096
+
+# Codemode
+codemode:
+  enabled: true
+
+# System prompt
+system_prompt: >-
+  {system_prompt}
+"""
+
+
+def yaml_list(items):
+    return "\n".join(f"  - {i}" for i in items)
+
+
+def main():
+    written = 0
+    for slug, (vertical, name, description, goal) in WORKERS.items():
+        cfg = VERTICALS[vertical]
+        tags = [vertical, "agent-worker"]
+        envvars_block = ""
+        if cfg["envvars"]:
+            envvars_block = "\n# Secrets / environment variables\nenvvars:\n" + yaml_list(cfg["envvars"]) + "\n"
+        system_prompt = (
+            f"You are the {name}, an autonomous agent worker in the "
+            f"{vertical.replace('-', ' ')} domain. {description} "
+            "Work step by step, show your reasoning and evidence, and require "
+            "explicit human approval before any external or irreversible action."
+        )
+        content = TEMPLATE.format(
+            name=name,
+            slug=slug,
+            vertical=vertical,
+            vertical_label=vertical.replace("-", " ").title(),
+            description=description,
+            tags=yaml_list(tags),
+            mcp=yaml_list(cfg["mcp"]),
+            skills=yaml_list(cfg["skills"]),
+            envvars=envvars_block,
+            icon=cfg["icon"],
+            emoji=cfg["emoji"],
+            color=cfg["color"],
+            goal=goal,
+            system_prompt=system_prompt,
+        )
+        path = AGENTS_DIR / f"workers-{slug}.yaml"
+        path.write_text(content, encoding="utf-8")
+        written += 1
+    print(f"Wrote {written} worker specs to {AGENTS_DIR}")
+
+
+if __name__ == "__main__":
+    main()

package/lib/examples/CellSimpleExample.d.ts

@@ -1,6 +0,0 @@
-import { ServiceManager } from '@jupyterlab/services';
-type IJupyterCellExampleProps = {
-    serviceManager?: ServiceManager.IManager;
-};
-export declare const JupyterCellExample: (props: IJupyterCellExampleProps) => import("react/jsx-runtime").JSX.Element;
-export default JupyterCellExample;

package/lib/examples/CellSimpleExample.js

@@ -1,20 +0,0 @@
-import { jsx as _jsx, jsxs as _jsxs } from "react/jsx-runtime";
-import { Cell, KernelIndicator, useJupyter, useKernelsStore, useCellsStore, } from '@datalayer/jupyter-react';
-import { ThemedJupyterProvider } from './utils/themedProvider';
-import { Button, Label } from '@primer/react';
-import { Box } from '@datalayer/primer-addons';
-const CELL_ID = 'cell-example-1';
-const DEFAULT_SOURCE = `from IPython.display import display
-
-for i in range(10):
-    display('I am a long string which is repeatedly added to the dom in separated divs: %d' % i)`;
-const JupyterCellExampleContent = () => {
-    const { defaultKernel } = useJupyter({ startDefaultKernel: true });
-    const cellsStore = useCellsStore();
-    const kernelsStore = useKernelsStore();
-    return (_jsxs(Box, { p: 4, children: [_jsx(Box, { as: "h1", children: "Jupyter Cell Example" }), _jsxs(Box, { children: ["Source: ", cellsStore.getSource(CELL_ID)] }), _jsxs(Box, { children: ["Outputs Count: ", cellsStore.getOutputsCount(CELL_ID)] }), _jsxs(Box, { children: ["Kernel State:", ' ', _jsx(Label, { children: defaultKernel && kernelsStore.getExecutionState(defaultKernel.id) })] }), _jsxs(Box, { children: ["Kernel Phase:", ' ', _jsx(Label, { children: defaultKernel && kernelsStore.getExecutionPhase(defaultKernel.id) })] }), _jsxs(Box, { display: "flex", children: [_jsx(Box, { children: "Kernel Indicator:" }), _jsx(Box, { ml: 3, children: _jsx(KernelIndicator, { kernel: defaultKernel && defaultKernel.connection }) })] }), _jsx(Box, { children: _jsx(Button, { onClick: () => cellsStore.execute(CELL_ID), children: "Run cell" }) }), _jsx(Cell, { source: DEFAULT_SOURCE, id: CELL_ID, kernel: defaultKernel })] }));
-};
-export const JupyterCellExample = (props) => {
-    return (_jsx(ThemedJupyterProvider, { children: _jsx(JupyterCellExampleContent, {}) }));
-};
-export default JupyterCellExample;