Zopaf Negotiation Engine

Negotiation math engine — Pareto frontiers, iso-utility counteroffers, and preference inference via MILP optimization. Zero LLM tokens.

What to check before installing or using this skill: - Understand the default data flow: SKILL.md/README point at https://zopaf-mcp-production.up.railway.app/mcp. By default, your negotiation terms, BATNA, and inferred priorities would be sent to that external server. If terms are sensitive (job offers, term sheets, M&A, procurement), do NOT use the default endpoint. - Prefer on-prem or trusted-host deployment: the repository includes zopaf_mcp_server.py and instructions to run the MCP server locally. If you need confidentiality, run the included server in your environment and set mcp.servers.zopaf to your local URL (openclaw mcp set zopaf ...). Review the server code (zopaf_mcp_server.py) and the web app before deploying. - Verify the "zero tokens" claim: the math engine offloads optimization, but the included coach/agent code calls LLM providers (anthropic/Claude). Expect LLM usage/tokens for the conversational side unless you replace that component or run a fully local LLM stack. - Audit network behavior and logging: inspect zopaf_mcp_server.py, coach.py, api.py and any web UI code for outgoing network calls, error reporting, or telemetry. Confirm whether logs or demo data are stored or forwarded to third parties (the repo references demo-data.json and a web UI). Look for any hard-coded endpoints. - Trust and provenance: source and homepage are 'unknown'/'none' and the registry metadata owner is not a known organization. If you must use the hosted endpoint, ask the maintainer/operator about their privacy/data-retention policy, who controls the Railway deployment, and whether data is stored, logged, or used to train models. - If you want to proceed safely: run the server locally, point mcp.servers.zopaf at the local instance, and optionally remove or audit any telemetry. If you cannot host locally, avoid sending sensitive negotiation data to the public endpoint and treat the package as untrusted until provenance is established.

Type: OpenClaw Skill Name: zopaf Version: 1.1.0 The Zopaf skill bundle is a comprehensive negotiation math engine and coaching tool. It uses Mixed-Integer Linear Programming (MILP) via the PuLP library to calculate Pareto frontiers and generate optimal deal structures. The bundle includes a FastAPI backend (api.py), a Next.js frontend (web/), and an MCP server implementation (mcp_api.py, zopaf_mcp_server.py) that exposes negotiation tools to AI agents. The code is well-structured, follows its stated purpose of providing quantitative negotiation analysis, and contains no evidence of malicious intent, data exfiltration, or unauthorized execution. External connections are limited to a hosted MCP endpoint on Railway and a standard Stripe payment link for the optional 'premium' strategy features.