io.github.fnobbe/crashtestyourstrategy

Est-ce que io.github.fnobbe/crashtestyourstrategy est sûr ?

Au dernier audit MCP Hub, le serveur MCP « io.github.fnobbe/crashtestyourstrategy » est classé « Non encore testé ». Référencé dans l'annuaire. Seul le pentest comportemental fait foi, et il n'a pas encore été exécuté sur ce serveur. Le verdict est réévalué en continu ; consultez la fiche pour l'état à jour.

Méthode : MCP Hub analyse la description de chaque outil (couche 1) et, pour le verdict fort, exécute le serveur dans une micro-VM isolée avec leurres et réseau bloqué (couche 2). On publie des noms et des compteurs, jamais le contenu brut renvoyé par un outil.

Les 16 outils

• backtest_integrity≈ 669 tokens

  Confront a backtest claim with its over-optimism failure modes before trusting it. Given an annualized Sharpe + the number of configurations tried + the backtest window (YYYY-MM-DD), returns: the DEFLATED Sharpe — the expected MAXIMUM Sharpe achievable by chance grows with the trial count, so a high in-sample Sharpe is a selection artifact (Bailey & López de Prado); which CRISIS REGIMES were ABSENT from the backtest window (untested, from the historical-anchor catalogue); and a base-rate caveat. If the trial count is unknown — the usual case for an agent reasoning from a backtest — the Sharpe is flagged as not-deflatable / UNPROVEN. All inputs optional; supply as many as known. Descriptive, not advisory.

  Paramètres, retour & exemple (8)

  Paramètres

  • kurt · number · optionnel
    Kurtosis of the strategy's returns (3 = normal).
  • skew · number · optionnel
    Skewness of the strategy's returns (0 = symmetric).
  • asset · ? · optionnel
    Asset context for the regime-coverage check (default: SPY as the equity-crisis reference).
  • n_trials · ? · optionnel
    Number of configurations tried before selecting this backtest — drives the deflated-Sharpe correction. Unknown → the claim is flagged UNPROVEN.
  • frequency · number · optionnel
    Return observations per year (252 = daily bars).
  • backtest_end · ? · optionnel
    Backtest window end (YYYY-MM-DD).
  • backtest_start · ? · optionnel
    Backtest window start (YYYY-MM-DD) — used to detect crisis regimes the window never contained.
  • annualized_sharpe · ? · optionnel
    The claimed annualized Sharpe ratio of the backtest.

  Exemple d'appel

  {"name": "backtest_integrity", "arguments": {"kurt": 1, "skew": 1, "asset": "texte", "n_trials": "texte", "frequency": 1, "backtest_end": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• challenge_strategy≈ 344 tokens

  Adversarial-evaluation primitive — the semantic integration layer of the platform. Given a strategy identifier, returns a 3-layer analysis: (1) outcome metrics in the worst regimes the strategy was evaluated against, (2) vulnerability profile in the 8-dimension strategy vulnerability ontology with severity classification, (3) descriptor attribution showing which regime descriptors most strongly couple to the strategy's failure. v1 supports only 'buy_and_hold' (the outcome matrix is built once per strategy); future versions will support arbitrary strategy specs once the parser-driven strategy backtest pipeline is wired in. Read ontology://strategy-vulnerabilities for the vulnerability vocabulary.

  Paramètres, retour & exemple (1)

  Paramètres

  • strategy_id · string · optionnel
    Strategy identifier; v1 supports only 'buy_and_hold'.

  Exemple d'appel

  {"name": "challenge_strategy", "arguments": {"strategy_id": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• describe_regime≈ 307 tokens

  Single-regime introspection: returns the median behavioural descriptors of a known regime, the z-scores vs the catalogue population (so you can see what makes THIS regime distinct from the average), an English characterisation generated from the most extreme descriptors, and the top 2 nearest neighbours as a preview. Complements find_similar_regime: that tool ranks neighbours of a target, this tool tells you what a single regime IS. Read this before searching if you want to reason about one regime first.

  Paramètres, retour & exemple (1)

  Paramètres

  • profile_hint · string · requis
    Synthetic stress-regime identifier, e.g. 'whipsaw_synthetic_spy'. Discover valid values via the regimes://available resource.

  Exemple d'appel

  {"name": "describe_regime", "arguments": {"profile_hint": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• factor_decomposition≈ 375 tokens

  Reveal HIDDEN risk concentration: a portfolio can be capital-diversified while its RISK is dominated by one factor. Returns the Euler risk-contribution decomposition (RC_i = w_i*(Sigma*w)_i / w'Sigma*w, summing to 1) alongside the capital weights, using the empirical covariance of real returns. For this universe each asset proxies a factor (SPY=equity-beta, TLT=duration, GOLD=real-asset, BTC=crypto). E.g. a 60/40 is ~83% equity risk; a 50/50 SPY/BTC is ~86% BTC risk despite 50/50 capital. Descriptive, not advisory.

  Paramètres, retour & exemple (1)

  Paramètres

  • holdings · array · requis
    Portfolio legs: list of {asset, weight} objects, e.g. [{'asset': 'SPY', 'weight': 0.6}, {'asset': 'TLT', 'weight': 0.4}]. Weights are normalised to sum to 1; assets must be in the substrate universe. Each substrate asset proxies a factor (SPY=equity beta, TLT=duration, GOLD=real asset, BTC=crypto).

  Exemple d'appel

  {"name": "factor_decomposition", "arguments": {"holdings": []}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• find_similar_regime≈ 498 tokens

  Nearest-neighbour retrieval over the cached regime catalogue. Provide EITHER a reference_profile_hint (use that bundle's median descriptors as target) OR a descriptor_target dict (partial spec, missing dimensions are ignored — only the provided ones contribute to distance). Optional asset_filter restricts to one asset. Returns top_n matches with similarity_score (0..1), euclidean distance in z-score space, and per-descriptor signed deltas so the agent can see WHY a regime matched. Read ontology://regime-descriptors for the descriptor definitions, and regimes://descriptors for the full catalogue.

  Paramètres, retour & exemple (4)

  Paramètres

  • top_n · integer · optionnel
    Number of nearest regimes to return.
  • asset_filter · ? · optionnel
    Restrict matches to one asset (e.g. 'SPY', 'BTC').
  • descriptor_target · ? · optionnel
    Partial target spec {descriptor_name: value}; only the provided dimensions contribute to the distance. Definitions: ontology://regime-descriptors.
  • reference_profile_hint · ? · optionnel
    Use this catalogue bundle's median descriptors as the search target (mutually exclusive with descriptor_target).

  Exemple d'appel

  {"name": "find_similar_regime", "arguments": {"top_n": 1, "asset_filter": "texte", "descriptor_target": "texte", "reference_profile_hint": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• get_dossier≈ 392 tokens

  Compile recorded diagnostic responses into ONE citable record — a proper process documents itself. Every envelope response (MCP and REST) is recorded automatically, keyed by its request_id. Provide explicit request_ids (compiled chronologically) or last_n for the most recent entries. Returns the entries with their gate signals (revision_required + grounding_summary each) plus a ready-to-cite markdown document; revision_required on the dossier itself flags workflows containing unaddressed gate signals. Single verbatim entries: GET /api/v1/dossier/{request_id} on the REST surface. A factual record, not an assessment — descriptive, never advisory.

  Paramètres, retour & exemple (2)

  Paramètres

  • last_n · integer · optionnel
    Alternatively: compile the N most recent recorded entries (ignored when request_ids is given).
  • request_ids · ? · optionnel
    Explicit request_ids to compile chronologically (take them from previous responses' request_id fields).

  Exemple d'appel

  {"name": "get_dossier", "arguments": {"last_n": 1, "request_ids": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• get_investment_thesis≈ 307 tokens

  Return the complete thesis for `slug`: the economic framework (pillars with [E]/[M]/[K] evidence grades, falsifiers and a deep-dive), the rule-based portfolio (asset blocks × conservative/balanced/offensive weights + sizing rationale), and the stress evidence (per-tier backtest, per-regime median drawdown, real historical episodes, pre-registered claim verdicts, and the hedge hold/break behaviour). This is the 'instant portfolio with all tested attributes'. Discover slugs with list_investment_theses(). Descriptive, not advisory — the agent decides suitability.

  Paramètres, retour & exemple (1)

  Paramètres

  • slug · string · requis
    Thesis slug — discover valid values via list_investment_theses().

  Exemple d'appel

  {"name": "get_investment_thesis", "arguments": {"slug": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• ips_gate≈ 520 tokens

  Check a portfolio against an Investment Policy Statement BEFORE accepting it — the planning step a proper process does FIRST (CFA). Provide holdings + IPS constraints (max_drawdown_tolerance as a fraction e.g. 0.15, time_horizon_years, liquidity_need 'low'|'medium'|'high'). Runs the stress test internally and flags where the proposal VIOLATES the stated policy: worst stress drawdown exceeds tolerance; a short horizon cannot absorb a deep drawdown; material holdings are less liquid than the stated need. A HARD GATE, not a score. Descriptive, not advisory.

  Paramètres, retour & exemple (4)

  Paramètres

  • holdings · array · requis
    Portfolio legs: list of {asset, weight} objects, e.g. [{'asset': 'SPY', 'weight': 0.6}, {'asset': 'TLT', 'weight': 0.4}]. Weights are normalised to sum to 1; assets must be in the substrate universe.
  • liquidity_need · ? · optionnel
    'low' | 'medium' | 'high' — violated when material holdings are less liquid than the stated need.
  • time_horizon_years · ? · optionnel
    Investment horizon stated in the IPS; short horizons cannot absorb deep drawdowns.
  • max_drawdown_tolerance · ? · optionnel
    IPS drawdown tolerance as a fraction, e.g. 0.15 = a -15% maximum acceptable drawdown.

  Exemple d'appel

  {"name": "ips_gate", "arguments": {"holdings": []}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• list_investment_theses≈ 258 tokens

  Discover the investment-thesis catalog. Each entry is a descriptive case study that pairs an economic framework with a rule-based portfolio and the synthetic + historical stress evidence for that allocation. Returns one compact summary per thesis (slug, title, one-liner, tags, risk tiers, framework summary, headline finding). Call get_investment_thesis(slug) for the full framework / portfolio / stress evidence, or read the thesis://{slug} resource. Descriptive, not advisory — the agent decides what is suitable.

  Paramètres, retour & exemple (0)

  Exemple d'appel

  {"name": "list_investment_theses", "arguments": {}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• long_horizon_stress≈ 878 tokens

  Distribution of multi-year wealth paths for a savings plan (monthly_contribution) or a withdrawal plan (monthly_withdrawal, inflation-indexed by default) on a portfolio from the substrate universe. Multi-year paths chain ~2y model blocks (block-bootstrap, disclosed); long-run drift is RE-ANCHORED to stated capital-market assumptions (overridable via long_run_drift; the substrate's raw stress drift would compound a structural bear universe — both are echoed in the output) while the model's path shape (vol, clustering, correlations, hedge-breaks) is kept. Costs are ON by default. Returns terminal-wealth quantiles (nominal + real), ruin/shortfall probabilities, a sequence-of-returns diagnosis (same plan, bad vs good first two years), and a drift-sensitivity block (assumptions − 2pp). Amounts in the caller's currency unit. Descriptive, not advisory — no rate, allocation, or product is recommended.

  Paramètres, retour & exemple (10)

  Paramètres

  • holdings · array · requis
    Portfolio legs: list of {asset, weight} objects, e.g. [{'asset': 'SPY', 'weight': 0.6}, {'asset': 'TLT', 'weight': 0.4}]. Weights are normalised to sum to 1; assets must be in the substrate universe.
  • rebalance · string · optionnel
    Rebalancing frequency: 'daily' | 'monthly' | 'quarterly'.
  • horizon_years · number · requis
    Plan horizon in years (multi-year paths are chained from ~2-year model blocks).
  • target_amount · ? · optionnel
    Optional wealth target; the output reports the probability of reaching it.
  • long_run_drift · ? · optionnel
    Override the re-anchored long-run drift per asset: {ASSET: annual drift fraction}; omit for the stated capital-market assumptions.
  • annual_inflation · number · optionnel
    Annual inflation assumption for indexing and real-value reporting (fraction, default 0.02).
  • initial_investment · number · optionnel
    Starting capital (account currency).
  • monthly_withdrawal · number · optionnel
    Monthly withdrawal (withdrawal-plan mode); inflation-indexed when withdrawal_inflation_indexed is true.
  • monthly_contribution · number · optionnel
    Fixed monthly savings contribution (savings-plan mode).
  • withdrawal_inflation_indexed · boolean · optionnel
    Index the monthly withdrawal to inflation.

  Exemple d'appel

  {"name": "long_horizon_stress", "arguments": {"holdings": [], "horizon_years": 1}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• market_regime_map≈ 416 tokens

  Compressed cross-category map of the current market state in ONE call: for 18 category proxies (US large-cap + tech, the 9 SPDR sectors, developed ex-US, emerging markets, long Treasuries, high-yield credit, gold, oil, Bitcoin) the operational regime (BULL/SIDEWAYS/BEAR/CRISIS), model-conditional regime probabilities over a 5- or 21-trading-day horizon, stress probability vs its unconditional baseline, a descriptive historical forward-return distribution conditional on the current regime label, and an equity-factor commonality flag (US sectors largely re-express one factor — the map is fewer independent signals than rows). Per (asset, horizon) cell only the preregistered, out-of-sample-validated model tier ships (covariate logit / persistence / unconditional — see tier_pvalues). Deliberately ships NO directional up/down forecast: regime membership is the validated signal, not return direction. Use regime_outlook for single-asset depth with as_of support. Descriptive, not a market prediction, not advisory.

  Paramètres, retour & exemple (1)

  Paramètres

  • horizon_days · integer · optionnel
    Validated horizons only: 5 or 21 trading days.

  Exemple d'appel

  {"name": "market_regime_map", "arguments": {"horizon_days": 1}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• portfolio_compare≈ 473 tokens

  Compare two portfolios (A = reference, B = candidate revision) on IDENTICAL simulated substrate paths — a paired design, so every delta is attributable to the weights, not seed noise. Returns drawdown-distribution deltas (median/worst/quantiles), probability-weighted scenario summaries, per-scenario outcome deltas, risk-concentration shift (Euler decomposition), and which diversification failures the candidate introduces or resolves. revision_required flags a candidate that deepens the worst-path drawdown or introduces a new diversification failure — the case where a revision made robustness worse. Provide holdings_a / holdings_b as lists of {asset, weight}. Descriptive, not advisory; neither portfolio is recommended or ranked.

  Paramètres, retour & exemple (2)

  Paramètres

  • holdings_a · array · requis
    Reference portfolio A. Portfolio legs: list of {asset, weight} objects, e.g. [{'asset': 'SPY', 'weight': 0.6}, {'asset': 'TLT', 'weight': 0.4}]. Weights are normalised to sum to 1; assets must be in the substrate universe.
  • holdings_b · array · requis
    Candidate revision B, same shape — evaluated on paths identical to A's, so every delta is attributable to the weights.

  Exemple d'appel

  {"name": "portfolio_compare", "arguments": {"holdings_a": [], "holdings_b": []}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• portfolio_stress_test≈ 585 tokens

  Stress a multi-asset portfolio across cross-asset regimes (baseline / risk_off_crisis / rate_shock). Provide `holdings` as a list of {asset, weight}; weights are normalised. Returns, per regime: portfolio return, worst-episode drawdown, a per-leg decomposition, and a cross_asset_finding (diversification_intact / hedge_holds / hedge_breaks / shared_drawdown) describing how the holdings behaved TOGETHER. The joint correlation structure (incl. the bond hedge that can break under rate shocks) is baked into a pre-computed substrate, so Tier-1 is instant over a fixed universe (read portfolio://universe). Optional `costs` ({rebalance: none|daily|monthly|quarterly|band, annual_costs: {asset: fraction}, transaction_cost_bps}) adds a cost_impact block: frictionless vs the stated rebalancing policy + costs via a path-loop engine with real unit accounting, paired on identical paths. The substrate is a fixed 4-asset universe (SPY, TLT, GOLD, BTC; read portfolio://universe). For ANY other ticker or a custom multi-asset book, use build_portfolio in assess mode (portfolios={name:{ticker:weight}}), which calibrates and stresses an arbitrary universe live. Descriptive, not advisory.

  Paramètres, retour & exemple (2)

  Paramètres

  • costs · ? · optionnel
    Optional cost model: {'rebalance': 'monthly', 'transaction_cost_bps': float, 'annual_costs': {ASSET: annual fraction}}. Omit for the frictionless default.
  • holdings · array · requis
    Portfolio legs: list of {asset, weight} objects, e.g. [{'asset': 'SPY', 'weight': 0.6}, {'asset': 'TLT', 'weight': 0.4}]. Weights are normalised to sum to 1; assets must be in the substrate universe.

  Exemple d'appel

  {"name": "portfolio_stress_test", "arguments": {"holdings": []}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• regime_outlook≈ 443 tokens

  Model-conditional probabilities that an asset is in each market regime (BULL / SIDEWAYS / BEAR / CRISIS, operational trailing-vol/drift labels) after a 5- or 21-trading-day horizon — the probability complement to the conditional stress tools: stress tools answer 'what happens GIVEN regime X', this answers 'how likely is regime X from today's observable state'. Ships only the preregistered, out-of-sample-validated tier (covariate logit; seasonality was tested and falsified); the persistence and unconditional baselines are reported alongside so an agent can see how much the model adds. Validated assets: SPY, QQQ, GLD, TLT. Optional as_of (YYYY-MM-DD) computes the outlook at a historical date. Probabilities describe membership in operationally defined regime classes — descriptive, not a market prediction, not advisory.

  Paramètres, retour & exemple (3)

  Paramètres

  • as_of · string · optionnel
    Optional historical evaluation date (YYYY-MM-DD); empty = latest data.
  • asset · string · optionnel
    One of the out-of-sample-validated assets: 'SPY', 'QQQ', 'GLD', 'TLT'.
  • horizon_days · integer · optionnel
    Validated horizons only: 5 or 21 trading days.

  Exemple d'appel

  {"name": "regime_outlook", "arguments": {"as_of": "texte", "asset": "texte", "horizon_days": 1}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• run_stress_test≈ 291 tokens

  Run a buy-and-hold backtest against the synthetic stress regime identified by profile_hint. Returns a structured diagnostic: robustness score (0-100), per-FM-bucket failure-behavior classification with confidence + context, and the resolved regime parameters that were actually evaluated. v1 supports only buy-and-hold. To discover available regime profile_hints, read the `regimes://available` resource. Diagnostic is descriptive, not advisory.

  Paramètres, retour & exemple (1)

  Paramètres

  • profile_hint · string · requis
    Synthetic stress-regime identifier, e.g. 'whipsaw_synthetic_spy'. Discover valid values via the regimes://available resource.

  Exemple d'appel

  {"name": "run_stress_test", "arguments": {"profile_hint": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

• submit_feedback≈ 561 tokens

  Persist structured improvement feedback about a previous tool response. Provide your agent identity, the request_id you are commenting on, and one or more feedback items each carrying category (from the FeedbackCategory ontology), severity, observation, optional suggested_action, and agent_confidence (0..1). Read `feedback://insights` to see aggregated cross-agent feedback.

  Paramètres, retour & exemple (7)

  Paramètres

  • agent_name · string · requis
    Your agent identity (model or product name).
  • request_id · ? · optionnel
    request_id of the response this feedback refers to.
  • agent_vendor · ? · optionnel
    Vendor of the submitting agent (e.g. 'Anthropic', 'OpenAI').
  • feedback_items · array · requis
    One or more items, each {category (FeedbackCategory ontology), severity, observation, suggested_action?, agent_confidence (0..1)}.
  • session_context · ? · optionnel
    Optional free-text context of the session/workflow the feedback arose in.
  • overall_confidence · number · requis
    Overall confidence in this feedback, 0..1.
  • platform_version_evaluated · ? · optionnel
    Schema/platform version the feedback refers to (e.g. 'ctys-agent-v1').

  Exemple d'appel

  {"name": "submit_feedback", "arguments": {"agent_name": "texte", "feedback_items": [], "overall_confidence": 1}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"result": {}}}

C'est votre serveur ?

Revendiquez-le sur l'espace créateur (preuve GitHub, domaine ou npm) pour afficher le badge « propriétaire vérifié » et lancer un pentest comportemental à la demande afin d'obtenir le badge « Vérifié ».

Créer un compte / revendiquer →

Source

https://mcp.crashtestyourstrategy.ai/mcp