- Home
- Skills
- Agents & Orchestration
- Aegis-World-Forge
More screenshots
Works with the AI Agents you already use
Aegis-World-Forge
L11 Causal World-Model Inductor with Z3 formal verification and 1,000-trajectory tail-stress simulation.
$99
Aegis-World-Forge
Example session with this skill installed
Induce a quant.liquidity world model from our orderbook transitions, reject any non-monotonic hallucinated impact law in Z3, run a 1,000-path counterfactual rollout during a 70% liquidity vacuum, and test CUSUM regime drift recalibration.
- Read your context and instructions
- Compiled the aegis-world-forge
I have executed the L11 Causal World-Model & Counterfactual Verification lifecycle across induce_and_verify_world_model, run_counterfactual_policy_rollout, and observe_reality_and_calibrate_drift.
Aegis-World-Forge L11 Verification & Simulation Audit
| L11 Stage / Scenario | Engine Verdict | Formal & Risk Guarantee | Latency / Impact |
|---|---|---|---|
| Hallucinated Negative-Slope Slippage Law | REJECTED_Z3_VIOLATION | Blocked NON_MONOTONIC_SLIPPAGE_LAW with SAT counterexample | 0 unsafe models |
Induced quant.liquidity World Model (v1) | WORLD_MODEL_ACTIVE | Capital conservation, depth >= 0, monotonicity & margin proven | <25 ms Z3 proof |
| 1,000-Path Rollout (Flash-Crash 70% Vacuum) | ABORT_LIQUIDITY_HALT | 0.0% tail liquidation (800 normal + 200 black-swan stress paths) | 4.68 ms (513x faster) |
Sustained Volatility Regime Shift (3 ticks) | WORLD_MODEL_ACTIVE (v2) | CUSUM triggered v1 -> v2 recalibration after hysteresis filter | <1 ms rollback ready |
Live Capital Protection & Efficiency Summary
Zero Tail Liquidations: 200 mandatory black-swan stress paths intercepted liquidity-vacuum traps before execution, reducing average realized slippage by 97.6% (9.63 bps vs 395.40 bps blind execution).
Extensible Architecture: Ships with quant.liquidity ready out of the box alongside extensible manifests for cloud.cluster and economy.virtual.
Next Steps
1.
Audit Custom Impact Equations: Call audit_custom_transition_law_z3 to formally verify proprietary slippage or state-transition formulas.
2.
Inspect Live Telemetry: Query get_world_forge_telemetry to track prevented liquidations, slippage saved in bps/USD, and Z3 proof counts.
Connects securely to your tools. The creator never sees your data.
About this skill
High-leverage automated trading systems often fail during flash crashes because their internal models don't account for tail risks or regime shifts. Aegis World Forge solves this by enforcing formal Z3 SMT theorem proving on all transition laws and running 1,000-path counterfactual digital twin simulations for every execution policy.
What it does
- Formal Verification proves four critical axioms including Capital Conservation and Margin Solvency via Z3 before activating any world model.
- Counterfactual Rollouts executes 1,000 Monte Carlo trajectories, specifically simulating 200 synthetic flash-crash vacuum paths with 70% depth drops.
- Regime Drift Calibration utilizes CUSUM error scores and hysteresis suppression to detect sustained market volatility shifts without thrashing on noise.
- AST Sandboxing enforces a strict execution barrier on custom transition laws to prevent unsafe imports or unbounded loops.
- Automated Liquidity Halts triggers immediate
ABORT_LIQUIDITY_HALTif tail-stress simulations yield non-zero liquidation probabilities.
How it works
- Use
induce_and_verify_world_modelto generate a Z3-verified causal world model from market observations. - Call
run_counterfactual_policy_rolloutto evaluate execution policies against normal and tail-stress trajectories. - Feed post-trade data into
observe_reality_and_calibrate_driftto monitor CUSUM regime drift and trigger auto-recalibration. - Execute
rollback_or_lock_world_modelfor sub-millisecond reverts between model generations if market conditions destabilize.
Frameworks & tools
This skill integrates Z3 SMT Theorem Prover, Qdrant vector indexing, and Python AST analysis. It is designed for high-frequency trading environments and quant research pipelines.
Why this beats prompting it yourself
Standard LLM prompts cannot guarantee physical monotonicity in liquidity curves or mathematically prove capital conservation. This skill replaces vague predictions with formal neuro-symbolic verification and rigorous tail-risk stress testing that LLMs cannot perform natively.
Use cases
- Preventing catastrophic liquidation during liquidity vacuums and flash crashes.
- Auditing custom slippage laws for mathematical consistency and safety.
- Automatically recalibrating trading models when market regimes shift volatility.
- Validating large-scale execution policies across hundreds of counterfactual paths.
Known limitations
Custom laws must be expressed as closed-form math to pass AST sandboxing. Induction requires a minimum of three consecutive out-of-regime ticks to trigger CUSUM recalibration.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
This software is a quantitative simulation and formal verification tool, not licensed financial or investment advice. Memory and calibrated world-model generations are local-first (volatile in-memory Qdrant) unless backed by persistent Qdrant storage. Transition laws must pass AST sandboxing and Z3 symbolic monotonicity/conservation proofs before activation, and counterfactual rollouts model parameterized microstructure regimes rather than live exchange order-routing execution
Custom laws must be closed-form math. Requires 3+ consecutive drift ticks for CUSUM trigger. Z3 proof time adds ~15ms latency.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed today
- Accounts
Qdrant
- Software
Python 3.11+z3-solver ( =4.12.0)qdrant-client ( =1.9.0)
- Optional
- SSE / HTTP transport mode (uses local stdio by default)
100% local-first L11 causal world-model compiler with zero external API key dependencies. Uses z3-solver for formal conservation/monotonicity proofs, embedded Qdrant for regime clustering, and deterministic pure-Python Monte Carlo rollouts for 4.68ms counterfactual policy evaluation.
Permissions
Allowed Hosts
File Scopes