mcp server · for AI agents

Give your AI agent a grumpy second opinion.

mostlyno speaks MCP. Any agent that can call tools — Claude, Cursor, your own bot — can query the council directly: verdicts, coverage, the disagreement board, the NO-ledger. Read-only, free, and every answer ships with its disclaimer baked in.

Connect

Claude Code / CLI:

claude mcp add --transport http mostlyno https://mostlyno.com/api/mcp

Claude Desktop, Cursor and other clients: add a remote MCP server with URL https://mostlyno.com/api/mcp (Streamable HTTP, no auth). Then try:

  • · “What does the mostly-no council think about PLTR?”
  • · “Where does the council disagree with Wall Street the hardest right now?”
  • · “Run my watchlist past mostlyno and tell me what it refuses to price.”

The tools

  • get_verdict

    One ticker → the council's stance in plain words, quality score, the honest reasons NOT to buy, and the conservative fair value — unless the name is too hard, in which case your agent gets the same refusal humans get.

  • search_coverage

    Search the ~1,500 covered S&P 1500 names by ticker or company name.

  • get_disagreements

    The Street-vs-council board: the biggest live fights with Wall Street consensus, plus the agreements, plus when the receipts unseal.

  • get_no_ledger

    Every refusal, counted: saves and misses vs the S&P 500, with the coin-flip control group.

Anonymous demo tier: 120 calls/hour per IP. API keys with higher quotas and the write-side tools (bot-league reporting, coverage requests) are on the roadmap — the data your agent gets is identical to the public pages, never earlier, never more.

What the data means: how we price · the disagreement board · the NO-ledger

Impersonal research & education only — not investment advice. The server speaks about securities, never about anyone's situation, and refuses to name prices on too-hard names — agents get the same honesty humans do.