Systematic token economic analysis. 51 risk checks, quantitative simulations, concrete fix proposals.
Most token models fail in predictable ways: emission shocks, death spirals, insider concentration, governance capture, treasury collapse. The patterns are well-documented. What's rare is finding them rigorously — before launch, with enough specificity to actually fix them.
This is a Claude Code agent that does exactly that. Submit your token model in any format and it runs a structured 7-step analysis: parsing, archetype classification, weakness scanning, quantitative simulation, and a full written report with grade and fix proposals.
| Deliverable | Description |
|---|---|
| Parsed TokenModel | Your model converted to structured YAML — a precise representation of every supply, vesting, governance, and treasury parameter |
| Completeness scorecard | Gap assessment before analysis begins; targeted follow-up questions for anything missing |
| Weakness scan | 51 checks across 5 severity levels, grounded in real failure postmortems; every Critical and High finding includes a specific, implementable fix proposal |
| Simulations | Supply curves, Monte Carlo price bands (1,000 paths), parameter sensitivity heatmaps, death spiral stress tests, viable operating region maps |
| Written report | A–F grade with justification, system dynamics analysis, equilibrium conditions, sustainability requirements, and optimization roadmap |
The analysis draws simultaneously from seven frameworks:
- Mechanism design — are the incentives actually compatible across all participant types?
- Evolutionary game theory — are the equilibria stable under adversarial pressure?
- Complex systems science — where are the phase transitions and tipping points?
- Institutional economics — does the governance design meet Ostrom's criteria for sustainable commons?
- Platform economics — how do network effects and bootstrapping dynamics interact with token incentives?
- Monetary theory — what are the seigniorage dynamics and monetary policy analogs?
- Financial economics — what are the embedded options and stochastic supply dynamics?
Findings, grades, and fix proposals are outputs of this analysis — not a checklist applied mechanically.
The agent loads domain knowledge selectively based on your token's features:
| File | Covers |
|---|---|
token_archetypes.md |
7 archetypes with classification decision tree and failure patterns |
failure_postmortems.md |
Terra/LUNA, Iron Finance, OHM, Anchor, Basis Cash; 10-condition death spiral checklist |
failure_postmortems_2024.md |
Stream Finance/xUSD, Compound governance capture, leverage stablecoin patterns |
staking_dynamics.md |
Issuance formulas, reflexivity loop, staking equilibrium model |
governance_attacks.md |
Beanstalk, Tornado Cash, Build Finance; 22-item governance attack checklist |
vetokens_and_emissions.md |
veCRV mechanics, Solidly failure, emission decay curves, bribe market dynamics |
treasury_design.md |
Runway formula, diversification benchmarks, protocol-owned liquidity |
fee_economics.md |
Fee capture ratio, revenue taxonomy, unit economics benchmarks |
token_velocity.md |
MV=PQ framework, velocity trap, sink mechanism effectiveness |
reference_benchmarks.md |
Live comparison data: BTC / ETH / UNI / CRV / GMX / stETH |
Supply composition over time with vesting cliff markers, showing freely circulating vs. locked supply and Base / Bear / Stress scenario comparison.
Price bands (P10–P90) across Base, Bear, and Stress scenarios over a 9-year horizon, surfacing dilution risk and emission-driven sell pressure windows.
Parameter space maps showing which combinations of design parameters keep the system stable (green) vs. drive it toward collapse (red). Used to derive concrete sustainability thresholds.
See SUBMISSION_GUIDE.md — any format is accepted (prose, bullet points, whitepaper section, spreadsheet). No form to fill out.
Prerequisites: Claude Code and Python 3.10+
pip install -r requirements.txtOpen a Claude Code session in this directory and use the slash commands:
/audit path/to/your-token-model.md # full pipeline: parse → scan → simulate → report
/audit "describe your token here" # same, from a verbal description
/assess-model "..." # completeness check only — scorecard + gap questions
/parse-model "..." # parse to structured YAML, no full audit
/simulate models/my-token.yaml # run simulations for an existing parsed model
/stress-test models/my-token.yaml # adversarial scenarios: bear market, whale exit, revenue collapse
/benchmark models/my-token.yaml # comparison table against comparable protocols
Outputs are saved to:
| Artifact | Location |
|---|---|
| Parsed model | models/<token-name>.yaml |
| Scan trace | analysis/<token-name>/scan_trace.md |
| Charts and CSVs | analysis/<token-name>/ |
| Report (markdown) | reports/<token-name>_audit.md |
| Report (PDF) | reports/<token-name>_audit.pdf |
Python: numpy, pandas, matplotlib, scipy, openpyxl, jinja2, weasyprint, pyyaml
Agent: Claude Code (claude-sonnet-4-6 or later)