SNHP
About
Free game-theory negotiation advisor for AI agents, plus paid receipted sessions and agent memory.
Details
- Author
- ryuxik
- Categories
- Finance
Jump to
Setup
Install SNHP in your MCP client (Claude Desktop, Cursor, Windsurf, and others).
Repository: https://github.com/ryuxik/snhp
Follow the installation instructions in the repository README, then restart your MCP client.
4. The build-vs-buy test: SNHP vs a STRONG production prompt
Numbers (2) and (3) above are vs ageneralvanilla prompt. The sharper question — "why not just prompt the LLM well?" — is answered by running SNHP against a strong production prompt (snhp/llm_strong_baseline.py, whose system prompt even includes logrolling advice). On the 4-issue contract, Haiku+SNHP-tool vs Haiku+strong-prompt, n=12 paired seeds (python -m snhp.strong_baseline_headtohead, result committed atgametheory/server/static/strong_baseline_headtohead.json):
SNHP beats even a strong production prompt — but by roughlyhalfthe edge it shows against a weak one. Caveats: n=12, Haiku (not Sonnet), one contract domain; re-run at larger n / a stronger model to tighten the CI.
Network effect: the cooperation premium requires both sides to be SNHP-staked. Asymmetric matchups (Sonnet+SNHP vs vanilla Sonnet) lose 0.11 utility vs symmetric scaffolded play. Peer-mode advisor only fires when counterparty has posted a verifiable SNHP attestation.
Live demo (replay of the actual API trace at seed=42):https://snhp.dev/demo.html
In the committed round-robin (leaderboard/results/leaderboard.json,n_rounds=20), SNHP's rank by average utility depends on the market:
So SNHP is #1in the asymmetric marketsandmid-pack in the symmetric one— do not read this as "#1 overall." Its variance is the smallest in the field. Atn_rounds=100the symmetric field restabilizes further and Aspiration leads.
This NegMAS agent (snhp/negmas_agent.py) is aresearch artifact and is NOT the shipped product recommender— the product claims below are measured on the shipped code, not on this tournament.
Seegametheory/evals/README.mdfor the eval/tuning runbook.
- Tier 1 — Negotiation: sell-side + buy-side recommenders, anchor-attack detection, cryptographic first-strike commit-reveal, LLM-drafted reply emails (paid).
- Tier 2 — Auctions: Vickrey / first-price BNE / English ascending, Myerson optimal reserve, format recommendation, MC simulation.
- Tier 3 — Mechanism design: Gale-Shapley, asymmetric Myerson optimal auction, Gallego-van Ryzin posted-price.
Tier 4 (coalition games) deferred until a paying buyer asks for it.
mcp-name: io.github.ryuxik/snhp-negotiation
Bridge Town is an MCP-native, git-versioned financial modeling platform for FP&A teams and finance leaders. AI agents use Bridge Town tools to create projects, write Python model files, run models in isolated cloud sandboxes, query data, write outputs to Google Sheets, create dashboards, branch scenarios, and collaborate with teammates.
The Capital.com MCP Server lets your AI assistant talk to your trading account directly. Market data, position checks, trade previews – all in plain language, without leaving your AI tool.
Coinrule Agentic Trading MCP enables investors to create, backtest, execute, and manage trading agents through natural language across stocks, crypto and ETFs
Invest with Claude and other AI assistants
Australian Consumer Data Right Product Data
Remote MCP server for historical crypto & prediction-market data: search ~500K instruments, live market stats (OHLC, turnover, spreads, depth, slippage) and tick-data purchase. Keyless for catalog & stats; optional OAuth for account tools. Endpoint: https://cryptostruct.com/mcp
Cross-border debt collection from your AI assistant: check cases, get pricing, submit new cases.
Read-only MCP server for your Evibe investment portfolio + live market data (holdings, performance, dividends, benchmarks, screeners). Works with Claude & ChatGPT.
Financial and quantitative modeling engine for AI agents. Typed, named, deterministic.
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
