TreLLM Showcase: Quickstart Tour

Run one command, open one page, and inspect the artifacts that demonstrate agent reliability, realistic execution, risk-aware action filters, diagnostic visuals, and extensible plugins. Each card names the artifact and the command that regenerates it; the first-run path uses deterministic agents, tracked snapshots, and no live provider calls.

First run: no provider key Execution realism Risk lifecycle Replayable trajectories Extensible plugins

3-Minute Demo Video

Watch the quickstart command, showcase portal, audit report, execution realism, extension walkthrough, and retail planning sandbox without leaving this page.

First-run quickstart portal

Start with a guided quickstart route through the main demo artifacts, their source commands, and redacted cache manifest.

python scripts/run_launch_demo.py

Artifact status: ready

Benchmark v0.2 card

A compact result page for agent reliability, execution-aware baselines, intraday portfolio probes, and representation robustness.

python scripts/build_benchmark_page.py

Artifact status: ready

TradeArena leaderboard registry

A redacted benchmark-submission page that compares runs without exposing raw provider prompts or responses.

python scripts/build_benchmark_registry.py examples/benchmark_submissions

Artifact status: ready

Experiment-design demos

Execution realism, Markowitz/MVO baselines, representation signatures, and custom plugin extensibility.

python scripts/run_paper_design_demos.py

Artifact status: ready

Animated visual tour

Regenerate the README animations and inspect what each preview conveys without relying on motion alone.

python examples/visual_tour_demo.py

Artifact status: ready

Audit report

Trace one decision from market observation through proposal, risk review, execution, and reflection.

python scripts/render_audit_report.py

Artifact status: ready

Agent Autopsy Dashboard

Inspect intent versus executed weights, slippage attribution, and the risk intervention timeline from a replayable trajectory.

python scripts/render_agent_autopsy_dashboard.py

Artifact status: ready

Crisis gallery

Representation trajectory, correlation/intent heatmap, feedback curves, and exposure waterfall snapshots.

python examples/crisis_snapshot_demo.py

Artifact status: ready

A-share rule stress

T+1, price-limit, and board-lot constraints as auditable risk-gate interventions.

python examples/ashare_market_rules_demo.py

Artifact status: ready

Crypto microstructure stress

No-key high-volatility crypto-style stress with fill, rejection, latency, and slippage diagnostics.

python examples/crypto_microstructure_stress_demo.py

Artifact status: ready

Futures roll risk

Contract metadata and a roll schedule produce expiry and roll-window risk reports.

python examples/futures_roll_risk_demo.py

Artifact status: ready

Mock deep-RL policy baseline

A deterministic policy wrapper emits normal decisions and reuses risk, execution, trajectory, and evaluator plugins.

python examples/rl_policy_baseline_demo.py

Artifact status: ready

Alpaca paper export

Approved orders become paper-review JSON/CSV rows without any live broker submission.

python examples/alpaca_paper_export_demo.py

Artifact status: ready

Dry-run broker adapter

Broker request shape is validated locally without credentials, network calls, or live submission.

python examples/dry_run_broker_adapter_demo.py

Artifact status: ready

Broker capability manifest

Adapter permissions, account modes, credential policy, and live-safety controls are declared before review.

python examples/broker_capability_manifest_demo.py

Artifact status: ready

Broker approval safety

A redacted approval artifact becomes a live-mode safety gate that allows bounded orders and blocks oversized ones.

python examples/broker_approval_safety_demo.py

Artifact status: ready

Broker response reconciliation

Paper broker responses are matched back to submitted client order IDs for audit review.

python examples/broker_response_reconciliation_demo.py

Artifact status: ready

Operator runbook checklist

Human-gated live-readiness controls are written as an offline checklist artifact.

python examples/operator_runbook_demo.py

Artifact status: ready

Live-readiness preflight bundle

Capability, handoff, approval binding, response reconciliation, and runbook artifacts are checked together.

python examples/live_readiness_preflight_demo.py

Artifact status: ready

Holdings CSV import

A tiny holdings CSV fixture feeds the retail planning sandbox and paper rebalance diagnostics.

python examples/holdings_csv_import_demo.py

Artifact status: ready

Custom plugin extension

A local analyst plugin running through the same strategy, risk, execution, memory, and evaluator stack.

python examples/custom_plugin_demo.py

Artifact status: ready

Contributor extension walkthrough

Swap in a custom analyst, risk manager, and evaluator while reusing the rest of the TreLLM stack.

python examples/extension_walkthrough_demo.py

Artifact status: ready

Retail planning sandbox

Review investor profiles, suitability checks, target allocations, futures margin estimates, and paper rebalance orders.

python examples/retail_planner_demo.py

Artifact status: ready