Developer Hub

Connect AI agents to the governed FinanceGPT tool surface using stateless MCP 2026-07-28.

Discover active governed Quantitative Language Model profiles available to the authenticated workspace.

Discover verified third-party APIs, models and datasets distributed through the FinanceGPT marketplace gateway.

Discover FinanceGPT MCP tools, required scopes and the current credential grant state.

Invoke an active FinanceGPT QLM with optional governed quantitative model runs.

Read the workspace compute class, region policy, contractual data grants, usage and deployment-truth posture.

Read normalized accounting snapshots from governed connected-data sources.

Read normalized accounting and bank transactions with pagination and provenance.

Search canonical instruments across provider identifier mappings.

Retrieve normalized licensed price bars for a canonical instrument.

Retrieve normalized company fundamental facts with reporting dates and provenance.

Discover licensed economic and macroeconomic time series.

Generate and retrieve explainable, backtested FinanceGPT forecasts.

Discover M28 workflow recipes with inputs, outputs, risk class and measurement profiles.

Install a governed workflow recipe into the authenticated workspace as an M6 workflow template.

Read workspace proof-of-value programmes and evidence-backed outcome metrics.

Access EquityGPT professional portfolio risk metrics and benchmark-relative analytics.

Read workspace-scoped company, valuation and ESG evidence with source and verification metadata.

Run deterministic long-only portfolio construction across minimum variance, risk parity and mean-variance candidates.

Run deterministic price and verified-sensitivity factor shocks against a governed portfolio.

Generate a proposal from verified tax lots and provider-backed prices without placing replacement trades.

Read published peer percentiles from consenting, threshold-protected cohorts.

Retrieve a compact cross-domain FinanceGPT intelligence summary for embedded products.

Run the approved FinanceGPT financial anomaly autoencoder against workspace-derived features.

Generate reproducible governed financial stress scenarios from an approved FinanceGPT VAE.

Discover governed point-in-time financial features with methodology and source requirements.

Retrieve API-releasable feature values for a canonical FinanceGPT instrument.

Generate a sealed point-in-time feature matrix for a governed project universe.

Retrieve a sealed research snapshot and its point-in-time feature matrix.

Run a governed feature-quality experiment against a sealed research snapshot.

Discover validated deterministic FinanceGPT quantitative models across valuation, portfolio, risk, fixed income, derivatives, performance, forecasting, econometrics, volatility and factor analytics.

Execute a validated deterministic FinanceGPT quantitative model with supplied inputs.

Discover FinanceGPT forecasting, econometrics, volatility, factor and statistical-risk models.

Backtest approved forecasting models on a holdout tail, rank by error, and refit the winner for the requested forward horizon.

Discover approved FinanceGPT generative financial models, model cards, algorithms and evaluation evidence.

Generate reproducible financial feature scenarios conditioned on contraction, neutral or expansion regimes.

Generate seeded multi-period synthetic financial stress paths from an approved time-series VAE.

List workspace backtest, simulation and optimization jobs with durable status and result hashes.

Queue a governed historical strategy backtest and poll the durable compute job for results.

Queue seeded bootstrap or Gaussian Monte Carlo simulations with percentile paths and terminal-risk metrics.

Queue a bounded parameter sweep against an approved FinanceGPT quantitative model and objective.

Resolve the authenticated FinanceGPT workspace and API environment.

Read one active QLM profile, its language runtime, quantitative grants and governance boundary.

Read hash-only MCP tool invocation audit metadata for the authenticated agent credential.

Run the deterministic governance-readiness evaluation for a QLM profile version.

Read immutable QLM version numbers and sealed configuration hashes.

Discover FinanceGPT compute service classes, concurrency, runtime, GPU eligibility and service-target metadata.

Discover configured FinanceGPT execution regions and their explicit deployment and health state.

Read 30-day compute units, queue/compute latency, estimated cost telemetry and region distribution.

Create a SHA-256 sealed lineage manifest for workspace compute, quantitative model or ML inference evidence.

Retrieve a sealed FinanceGPT lineage manifest by export UID.

Read workspace-specific contractual data-use overlays by dataset, environment and region.

Discover FinanceGPT datasets that are currently licensed for API redistribution.

Resolve a canonical instrument with issuer, venue and licensed identifiers.

Read governed FinanceGPT forecast revisions and their backtest/evaluation metadata.

Retrieve licensed dividends, splits and other normalized corporate actions.

Resolve one canonical economic series and its provider provenance.

Retrieve economic observations with vintages for look-ahead-safe research.

Read governed recurring finance workflow and reporting-pack status.

Create a governed 7-90 day proof-of-value programme around selected workflow recipes.

Read source-referenced tax lots available to governed tax-aware portfolio analysis.

Read approval-gated investment reporting evidence.

Discover approved active FinanceGPT quantitative and generative model versions.

Resolve one governed financial feature definition.

Read workspace-scoped quantitative research projects and their governed feature sets.

Resolve one FinanceGPT quantitative model, version, methodology, schemas and validation evidence.

Read workspace-scoped quantitative model execution history, hashes and model versions.

List workspace forecast-model tournaments and their holdout evaluation evidence.

Retrieve one governed forecast model tournament, candidate errors and selected forward forecast.

Read workspace-scoped generative inference history, seeds, conditions, model versions and reproducibility hashes.

Retrieve one reproducible generative scenario or financial path run.

Retrieve one workspace-scoped compute job and its governed result payload.

Cancel a queued compute job before processing begins.