← Files Allocator AgentARCHIVED FILE
plugin.json
7.14 KB · Nov 2, 2026 · 13:15 UTC
{
"$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
"name": "c8-allocator-agent",
"version": "1.0.2",
"description": "Portfolio construction and risk analytics for investment teams: target weights under drawdown, turnover and band constraints, scenario comparison, rebalancing and walk-forward backtests.",
"author": {
"name": "C8 Index Technologies (EU) Limited",
"email": "support@c8-technologies.com",
"url": "https://c8-technologies.com"
},
"homepage": "https://c8-technologies.com/allocator-agent/support",
"keywords": [
"portfolio construction",
"asset allocation",
"risk parity",
"drawdown",
"backtest"
],
"extensions": {
"com.openai": {
"interface": {
"displayName": "Allocator Agent",
"shortDescription": "Portfolio construction & risk",
"longDescription": "Allocator Agent brings C8's portfolio-construction engine into ChatGPT. Name the funds you are considering - tickers, ISINs or fund names; their daily history comes from C8's market data - or give it your own return or NAV series, and it computes what the target allocation should be under the constraints you set: per-asset bounds, asset-class bands, a maximum drawdown limit (conditional drawdown at risk, 99%), a turnover cap from your current weights, and a minimum or maximum number of holdings.\n\nWhat you can do:\n- Build a model portfolio with one of fourteen construction methods, from inverse volatility and risk parity to constrained mean-variance and tactical allocation, and see which constraints bind.\n- Compare two to five set-ups side by side: what a drawdown cap costs in volatility, what a band changes, what a different method would have held.\n- Turn a target into a trade list from your current holdings, within a turnover limit.\n- Decompose a portfolio's risk: each holding's contribution, the same by asset class, the gap to your risk budgets, drawdown and concentration.\n- Evaluate performance over a period, or run a walk-forward backtest in which the portfolio is rebuilt at each date from the data available on that date only.\n\nEvery figure states its source - C8's validated engine, or arithmetic on top of it - and its basis. If a constraint set cannot be met, the answer says so and how close the engine got.\n\nAllocator Agent is decision support for asset managers, wealth managers and investment teams. It works on public price history and the data you provide: it does not give personal investment advice, forecast returns, execute trades or hold client funds. Uploaded series and results are kept for 24 hours so that follow-up questions work, then deleted; holdings quantities and portfolio values are never stored.\n\nNo tool changes or deletes your data; results are kept for 24 hours as described above. Requires a C8 Allocator Agent account. You connect with your account email and a one-time code; no API keys.",
"developerName": "C8 Index Technologies (EU) Limited",
"category": "Finance",
"capabilities": [
"Portfolio construction",
"Scenario comparison",
"Rebalancing to target",
"Risk decomposition",
"Performance evaluation",
"Walk-forward backtests"
],
"websiteURL": "https://c8-technologies.com",
"supportURL": "https://c8-technologies.com/allocator-agent/support",
"privacyPolicyURL": "https://c8-technologies.com/allocator_privacy.html",
"termsOfServiceURL": "https://c8-technologies.com/allocator_terms_of_use.html",
"defaultPrompt": [
"Build a risk parity portfolio from VWRL.LSE, AGGG.LSE and IGLN.LSE in GBP.",
"Compare risk parity and inverse volatility on the same three funds.",
"What portfolio construction methods can you run?"
],
"composerIcon": "./assets/icon.png",
"logo": "./assets/logo.png"
},
"review": {
"test_cases": {
"positive": [
{
"description": "Build a portfolio from named funds",
"prompt": "Build a risk parity portfolio from VWRL.LSE, AGGG.LSE and IGLN.LSE in GBP.",
"tools_triggered": "build_model_portfolio",
"expected_behavior": "Target weights for the three funds with ex-ante volatility, drawdown and each fund's risk share, stating the data window and that the series come from C8's market data."
},
{
"description": "Compare two construction methods",
"prompt": "Compare risk parity and inverse volatility on the same three funds.",
"tools_triggered": "compare_scenarios",
"expected_behavior": "Both portfolios side by side: weights, volatility, drawdown and turnover between them."
},
{
"description": "List the construction methods",
"prompt": "What portfolio construction methods can you run?",
"tools_triggered": "list_construction_methods",
"expected_behavior": "The fourteen methods with the inputs and constraints each accepts."
},
{
"description": "Risk decomposition of stated weights",
"prompt": "Where does the risk come from in 60% VWRL.LSE, 30% AGGG.LSE and 10% IGLN.LSE, in GBP?",
"tools_triggered": "analyze_risk",
"expected_behavior": "Each fund's contribution to portfolio volatility, the same by asset class, and the portfolio's drawdown and concentration."
},
{
"description": "Trades to reach a target within a turnover cap",
"prompt": "I hold 40% VWRL.LSE, 40% AGGG.LSE and 20% IGLN.LSE. What trades get me to a risk parity portfolio of the same funds if I can turn over at most 15%?",
"tools_triggered": "build_model_portfolio, rebalance_to_target",
"expected_behavior": "A trade list in weights that respects the 15% one-way turnover cap, with the resulting weights and the gap that remains to the target."
}
],
"negative": [
{
"description": "Return forecasts are out of scope",
"prompt": "Which of these funds will go up the most next month?"
},
{
"description": "Trade execution is out of scope",
"prompt": "Place these trades with my broker."
},
{
"description": "Personal suitability advice is out of scope",
"prompt": "Is this portfolio suitable for my client who retires next year?"
}
]
},
"demo_recording_url": "https://youtu.be/uNhtjpyra4A",
"commerce": false,
"commerce_description": "The plugin does not sell anything or take payments in ChatGPT. It requires a C8 Allocator Agent subscription bought on c8-technologies.com or in Microsoft Marketplace."
},
"publication": {
"release_notes": "First release: portfolio construction with fourteen methods and drawdown, turnover and band constraints, scenario comparison, rebalancing to target, risk decomposition, performance evaluation and walk-forward backtests."
},
"apps": "./.app.json"
}
}
}
SHA-256: 803fd0bb52e113ce6608f5da232f3ace16874ab5b5a97dcabc8c8429c06158cf