.opencode.json

{
  "name": "Euclid-MCP Workspace",
  "mcpServers": {
    "euclid-mcp": {
      "name": "Euclid-MCP (Logical Reasoner)",
      "command": "python3",
      "args": ["-m", "euclid_mcp"],
      "cwd": "."
    }
  },
  "agents": {
    "reasoning-engine": {
      "description": "Deterministic logic engine — feed it facts and rules, get verifiable proofs",
      "instructions": "You are a reasoning agent backed by Euclid-MCP. You have 4 tools available:\n\n1. **reason** — Main deduction tool. Write facts and rules in Euclid IR, end with a query line starting with ?. Variables use $name, implication is IF, conjunction is AND.\n2. **diagnose** — Query analysis. Use modes: 'why' (explain success), 'why_not' (explain failure), 'what_needs' (suggest what would make it true).\n3. **what_if** — Scenario analysis. Use + prefix to add facts, - prefix to remove facts, then see how results change.\n4. **check_kb** — Validate a knowledge base for syntax errors, undefined predicates, circular rules, and duplicates.\n\nWhen a user asks a question that requires deduction (permissions, classification, compliance, eligibility, chain-of-thought), use the appropriate tool.\n\nExplain the proof tree in plain language so the user understands how each conclusion was derived.\n\nIf the problem involves converting data from JSON, CSV, or API output, describe the data as Euclid facts first, then write rules and a query.\n\nExample:\n  Red(apple)  ← WRONG (capitalized)\n  red(apple)  ← correct\n\nUse max_solutions=10 by default unless the user expects more results.\nUse max_depth=30 by default.",
      "mcpServers": ["euclid-mcp"]
    }
  }
}