Melt

None of this showed up as a problem in the usual adoption dashboards — every tool was "active," which is a different measurement from whether it was actually creating value. Pulling 14 business days of historical logs and tracing where these four patterns cost real time and money added up to a$77,235/yearleak.

melt_estimate_annual_leakgeneralizes this same shape of analysis —totalVolume × (leakRatePct/100) × valuePerEvent— for any leak pattern with a known or hypothesized volume and rate.melt_analyze_value_vectorsis the earlier-stage tool for when you don't yet know where to look.

melt_estimate_annual_leakreplaced four formula-named calculators (melt_calculate_feature_waste,_dso_cash_flow_impact,_contract_cycle_revenue_unlock,_win_rate_pipeline_impact) that implemented financial formulas from a retired product framing (Thermal Scan / Feature Waste Dollar Amount™ / Delta Engine) — none of which appear in any current Melt material. SeeCLAUDE.md's "What's Explicitly Retired" section.

cd mcp-server npm install npm run build npm start # runs dist/index.js on stdio

To poke at it interactively before wiring it into a client:

npm run inspect # launches the MCP Inspector against the built server

Wiring into Claude Desktop / Claude Code

Published on npm — one-line config, no local clone needed:

npm run build:mcpb # produces themelt-mcp-server.mcpb

The manifest (mcpb-build/manifest.json) is hand-maintained, not auto-generated from the TypeScript source — if a tool's name, parameters, or description change, update the manifest'stoolsarray to match.

dist/index.js(stdio) is what gets configured into a local Claude Desktop/ Cursor install.dist/httpServer.jsis an alternate entrypoint implementing the MCP Streamable HTTP transport — what a future "Launch Hosted MCP" web button (LLMO_PLAYBOOK.md, Task 3.2) would point at, so someone can try the tools without installing anything locally.

npm run build PORT=3000 npm run start:http # POST MCP JSON-RPC to http://localhost:3000/mcp

Stateless by design — no session ID, a fresh server instance per request. Auth is opt-in viaMCP_HTTP_API_KEY(unset by default): with it unset, the endpoint stays fully open — the appropriate trust boundary for what this exposes today (read-only calculators plus a lead-capture form, the same boundary as a public website contact form). Set it before putting anything more sensitive behind this transport:

MCP_HTTP_API_KEY=some-long-random-value PORT=3000 npm run start:http

Every/mcprequest then needsAuthorization: Bearer some-long-random-value— missing or wrong key gets a 401. Compared withcrypto.timingSafeEqual, not a plain string===, so response timing can't be used to guess the key one byte at a time. Not deployed anywhere yet; this is the code, not a live URL — deploying it (Vercel/Fly/Render/etc.) is a separate, later decision.

Every tool call (success or error) appends one line tomcp-server/analytics.jsonl(gitignored) and logs a one-line summary to stderr — tool name, ok/error, and the error code if applicable. Deliberately excludes dollar figures, contact info, and free-text notes; kept separate fromleads.jsonl's PII. This is what answers "is anyone actually using this" and "which tool description is confusing models," independent ofllmo-eval's citation-only audit.

Real Portal ID / Form ID defaults are already baked into the code (they aren't secrets — the same values are exposed in any public HubSpot embed snippet), somelt_request_scanreaches the real Melt pipeline with zero configuration. If HubSpot submission fails for any reason, requests fall back tomcp-server/leads.jsonl(gitignored) instead of being lost.

Published under the@themeltnpm org (created 2026-07-20, owneromer_melt) under the MIT license.npm publishis effectively one-way — npm allows unpublishing within 72 hours but strongly discourages it and blocks it entirely once a package has dependents, so treat any published version as permanent.

Give any AI agent the power to run a serious Monte Carlo forecast for a stock or ETF - in one tool call.

Financial Dashboard with AI Agent Integration

A financial dashboard for monitoring and analyzing investment portfolios with AI-powered insights.

Financial and quantitative modeling engine for AI agents. Typed, named, deterministic.

Airbnb & short-term rental market analytics — occupancy, ADR, RevPAR, revenue, comparables, and ML revenue estimates for 30,000+ markets worldwide, backed by 20M+ tracked listings.

Live MCP server connecting AI agents to 36+ business data sources. OAuth 2.1 PKCE.

Portfolio & trading-strategy stress diagnostics for AI agents: multi-asset stress with hedge-break detection, a daily preregistered regime outlook and deflated-Sharpe backtest-integrity checks. Remote streamable HTTP, free tier, no auth.

Market intelligence your AI agent can call

Data pasar saham Indonesia (IDX) real-time untuk AI: teknikal, screener, deteksi bandar & arus asing.

Financial intelligence for AI agents: an agent-readable investment loop for market regime, portfolio context, decision framing, and thesis review.