opencode.jsonc

{
  // Local patched schema: the published https://opencode.ai/config.json still describes V1
  // (no `mcp.servers`, no `disabled`) and flags this V2 config with schema "type errors".
  // JSON schema diagnostics cannot be suppressed with lint comments — patching the schema is
  // the only in-repo fix. Revert this line + delete schema/opencode.config.json once upstream
  // ships the V2 shape (see x-source/x-patched in that file).
  "$schema": "./schema/opencode.config.json",
  "autoupdate": true,
  "default_agent": "conductor",
  "instructions": [
    "{env:OPENCODE_SRC_ROUTE}/opencode/instructions/subagent-routing.md",
    "{env:OPENCODE_SRC_ROUTE}/opencode/instructions/verification-gate.md",
    "{env:OPENCODE_SRC_ROUTE}/opencode/instructions/serena.md",
  ],
  "agent": {
    "architect": {
      "description": "MUST delegate for architecture, system design, scalability, cross-module boundaries, or high-impact technical decisions.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PLANNER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/architect.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_PRIMARY}",
      "tools": {
        "bash": true,
        "edit": false,
        "read": true,
        "write": false,
      },
    },
    "ask": {
      "description": "General-purpose Q&A agent. Answers questions about the editor, project, technologies (Context7), research topics, or plan improvements. Read-only — delegates file writes to writer/researcher.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PLANNER}",
      "permission": {
        "task": {
          "*": "deny",
          "planner": "allow",
          "researcher": "allow",
          "writer": "allow",
        },
      },
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/ask.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_PRIMARY}",
      "tools": {
        "bash": true,
        "edit": false,
        "glob": true,
        "grep": true,
        "read": true,
        "task": true,
        "webfetch": true,
        "websearch": true,
        "write": false,
      },
    },
    "build": {
      "disable": true,
    },
    "build-error-resolver": {
      "description": "MUST delegate when build, typecheck, lint, or TypeScript errors occur. Fixes only those errors with minimal diffs.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PROGRAMMER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/build-error-resolver.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "write": true,
      },
    },
    "coder": {
      "description": "MUST delegate for any non-test implementation work: writing new code, applying spec, scaffolding modules, porting code, implementing fixes. Pure implementation only — caller owns planning, architecture, and review.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PROGRAMMER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/coder.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "write": true,
      },
    },
    "conductor": {
      "description": "Primary orchestrator. Routes every task to the matching specialist subagent. CANNOT write or edit files directly — all writes go through writer (docs) or coder (code).",
      "mode": "primary",
      "model": "{env:OPENCODE_MODEL_AGENTIC}",
      "permission": {
        "bash": "allow",
        "edit": "deny",
        "read": "allow",
        "task": {
          "*": "deny",
          "architect": "allow",
          "ask": "allow",
          "build-error-resolver": "allow",
          "coder": "allow",
          "conductor": "deny",
          "database-reviewer": "allow",
          "debt-cleanup": "allow",
          "doc-updater": "allow",
          "e2e-runner": "allow",
          "git-specialist": "allow",
          "planner": "allow",
          "researcher": "allow",
          "reviewer": "allow",
          "security-reviewer": "allow",
          "tdd-guide": "allow",
          "writer": "allow",
        },
        "write": "deny",
      },
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/conductor.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "*": true,
        "edit": false,
        "write": false,
      },
    },
    "database-reviewer": {
      "description": "MUST delegate for PostgreSQL, SQL, Supabase, RLS, migrations, schema design, query performance, or DB security.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PROGRAMMER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/database-reviewer.txt",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "write": true,
      },
    },
    "debt-cleanup": {
      "description": "Fixes technical debt from completed tasks — dead code, stale exports, config drift, untracked files, duplicates.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PROGRAMMER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/skills/debt-cleanup/SKILL.md",
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "write": true,
      },
    },
    "doc-updater": {
      "description": "MUST delegate for documentation, README updates, architecture notes, and codemaps. Keeps docs aligned with code changes.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PLANNER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/doc-updater.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "write": true,
      },
    },
    "e2e-runner": {
      "description": "MUST delegate for browser automation, Playwright work, and critical user-flow E2E tests. Generates, runs, and reports artifacts.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PLANNER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/e2e-runner.txt",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "write": true,
      },
    },
    "git-specialist": {
      "description": "MUST delegate for branch names, commits, staging, pushes, pull request creation/status, and git hygiene. Performs git-only work safely.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_MINI}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/git-specialist.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_TERTIARY}",
      "tools": {
        "bash": true,
        "edit": false,
        "glob": false,
        "grep": false,
        "read": true,
        "write": false,
      },
    },
    "plan": {
      "disable": true,
    },
    "planner": {
      "description": "MUST delegate for complex feature planning, implementation plans, large refactors, or unclear execution order. Produces plans only.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PLANNER}",
      "permission": {
        "task": {
          "*": "deny",
          "writer": "allow",
        },
      },
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/planner.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_PRIMARY}",
      "tools": {
        "bash": true,
        "edit": false,
        "read": true,
        "task": true,
        "write": false,
      },
    },
    "researcher": {
      "description": "MUST delegate for research, investigate topics, compare technologies, multi-source analysis. Read-only for codebase; writes comparison artifacts to .opencode/thoughts/comparisons/.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PLANNER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/researcher.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_PRIMARY}",
      "tools": {
        "bash": true,
        "edit": false,
        "glob": true,
        "grep": true,
        "read": true,
        "webfetch": true,
        "websearch": true,
        "write": true,
      },
    },
    "reviewer": {
      "description": "MUST delegate for code review, PR review, pre-merge review. Reviews diffs for bugs, security, quality, tests. Read-only — findings only; with --code, delegates implementation to coder (TDD RED→GREEN).",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PROGRAMMER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/reviewer.txt",
      "permission": {
        "task": {
          "*": "deny",
          "coder": "allow",
        },
      },
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "bash": true,
        "edit": false,
        "glob": true,
        "grep": true,
        "read": true,
        "task": true,
        "write": false,
      },
    },
    "security-reviewer": {
      "description": "MUST delegate for auth, user input, API endpoints, secrets, payments, PII, or sensitive data. Read-only — reports vulnerabilities; orchestrator dispatches fixes to coder.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PROGRAMMER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/security-reviewer.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "bash": true,
        "edit": false,
        "glob": true,
        "grep": true,
        "read": true,
        "write": false,
      },
    },
    "tdd-guide": {
      "description": "MUST delegate for new features, bug fixes, or refactors that need tests. Enforces RED-GREEN-REFACTOR and 80%+ coverage. Writes test files only; delegates implementation to coder.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PROGRAMMER}",
      "permission": {
        "task": {
          "*": "deny",
          "coder": "allow",
        },
      },
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/tdd-guide.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_SECONDARY}",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "task": true,
        "write": true,
      },
    },
    "title": {
      "description": "Auto-generate session titles from first user message (enabled by default)",
      "maxTokens": 80,
      "model": "{env:OPENCODE_MODEL_SUBAGENT_MINI}",
    },
    "writer": {
      "description": "MUST delegate for writing or editing non-code text artifacts: documentation, README, markdown, HTML reports, release notes, ADRs, prose, change logs. Forbidden from touching source code files.",
      "mode": "subagent",
      "model": "{env:OPENCODE_MODEL_SUBAGENT_PLANNER}",
      "prompt": "{env:OPENCODE_SRC_ROUTE}/opencode/prompts/agents/writer.txt",
      "reasoningEffort": "{env:OPENCODE_REASONING_TERTIARY}",
      "tools": {
        "bash": true,
        "edit": true,
        "glob": true,
        "grep": true,
        "read": true,
        "write": true,
      },
    },
  },
  "command": {
    "ask": {
      "agent": "ask",
      "description": "Ask questions about the project, editor, technologies, or improve plans",
      "subtask": true,
      "template": "{file:commands/ask.md}\n\n$ARGUMENTS",
    },
    "build-fix": {
      "agent": "build-error-resolver",
      "description": "Fix build and TypeScript errors with minimal changes",
      "subtask": true,
      "template": "{file:commands/build-fix.md}\n\n$ARGUMENTS",
    },
    "discuss": {
      "description": "Collaborative decision-making with source citations and tradeoff analysis",
      "template": "{file:commands/discuss.md}\n\n$ARGUMENTS",
    },
    "fix": {
      "agent": "debt-cleanup",
      "description": "Clean up technical debt from the last task — dead code, broken exports, config drift, stale refs, untracked files",
      "subtask": true,
      "template": "{file:commands/fix.md}\n\n$ARGUMENTS",
    },
    "git": {
      "agent": "git-specialist",
      "description": "Handle git operations with enforced branch and commit conventions",
      "subtask": true,
      "template": "{file:commands/git.md}\n\n$ARGUMENTS",
    },
    "git-workflow": {
      "agent": "git-specialist",
      "description": "Full git workflow with PR stop-gate. /git-workflow bcps <branch>, bscps <branch>, or cps (default)",
      "subtask": true,
      "template": "{file:commands/git-workflow.md}\n\n$ARGUMENTS",
    },
    "learn": {
      "description": "Review session or git context to improve settings and memory",
      "template": "{file:commands/learn.md}\n\n$ARGUMENTS",
    },
    "opsx:apply": {
      "description": "Implement tasks from an OpenSpec change (Experimental)",
      "template": ".opencode/references/opsx/opsx-apply.md",
    },
    "opsx:archive": {
      "description": "Archive a completed change in the experimental workflow",
      "template": ".opencode/references/opsx/opsx-archive.md",
    },
    "opsx:bulk-archive": {
      "description": "Archive multiple completed changes at once",
      "template": ".opencode/references/opsx/opsx-bulk-archive.md",
    },
    "opsx:continue": {
      "description": "Continue working on a change - create the next artifact (Experimental)",
      "template": ".opencode/references/opsx/opsx-continue.md",
    },
    "opsx:explore": {
      "description": "Enter explore mode - think through ideas, investigate problems, clarify requirements",
      "template": ".opencode/references/opsx/opsx-explore.md",
    },
    "opsx:ff": {
      "description": "Create a change and generate all artifacts needed for implementation in one go",
      "template": ".opencode/references/opsx/opsx-ff.md",
    },
    "opsx:new": {
      "description": "Start a new change using the experimental artifact workflow (OPSX)",
      "template": ".opencode/references/opsx/opsx-new.md",
    },
    "opsx:onboard": {
      "description": "Guided onboarding - walk through a complete OpenSpec workflow cycle with narration",
      "template": ".opencode/references/opsx/opsx-onboard.md",
    },
    "opsx:propose": {
      "description": "Propose a new change - create it and generate all artifacts in one step",
      "template": ".opencode/references/opsx/opsx-propose.md",
    },
    "opsx:sync": {
      "description": "Sync delta specs from a change to main specs",
      "template": ".opencode/references/opsx/opsx-sync.md",
    },
    "opsx:update": {
      "description": "Update a change - revise existing planning artifacts and keep them coherent (Experimental)",
      "template": ".opencode/references/opsx/opsx-update.md",
    },
    "opsx:verify": {
      "description": "Verify implementation matches change artifacts before archiving",
      "template": ".opencode/references/opsx/opsx-verify.md",
    },
    "plan": {
      "agent": "planner",
      "description": "Create a detailed implementation plan for complex features",
      "subtask": true,
      "template": "{file:commands/plan.md}\n\n$ARGUMENTS",
    },
    "research": {
      "agent": "researcher",
      "description": "Multi-mode research: compare options or deep-dive single topics. Uses --compare/--deep flags.",
      "subtask": true,
      "template": "{file:commands/research.md}\n\n$ARGUMENTS",
    },
    "review": {
      "agent": "reviewer",
      "description": "Review code, PRs, or documentation — /review [target] [--docs] [--no-tools]",
      "subtask": true,
      "template": "{file:commands/review.md}\n\n$ARGUMENTS",
    },
    "security": {
      "agent": "security-reviewer",
      "description": "Run comprehensive security review",
      "subtask": true,
      "template": "{file:commands/security.md}\n\n$ARGUMENTS",
    },
    "skill-plus": {
      "description": "Create, improve, or generate skills from history — /skill-plus --create|--from-history|--improve",
      "template": "{file:commands/skill-plus.md}\n\n$ARGUMENTS",
    },
    "tune": {
      "description": "Tune opencode settings — commands, instructions, or config. --local/--global scope. Invokes customize-opencode skill; delegates writes to coder.",
      "template": "{file:commands/tune.md}\n\n$ARGUMENTS",
    },
    "update-codemaps": {
      "agent": "doc-updater",
      "description": "Update codemaps",
      "subtask": true,
      "template": "{file:commands/update-codemaps.md}\n\n$ARGUMENTS",
    },
    "update-docs": {
      "agent": "doc-updater",
      "description": "Update documentation",
      "subtask": true,
      "template": "{file:commands/update-docs.md}\n\n$ARGUMENTS",
    },
  },

  "mcp": {
    "servers": {
      "Figma": {
        "disabled": true,
        "headers": {},
        "type": "remote",
        "url": "https://mcp.figma.com/mcp",
      },
      "context7": {
        "command": ["{env:CONTEXT7_MCP_BIN}"],
        "type": "local",
      },
      "serena": {
        "command": ["serena", "start-mcp-server", "--context=ide"],
        "type": "local",
      },
    },
  },
  "plugin": [
    "opencode-skill-creator",
    "opencode-agent-memory",
    "model-fallback",
  ],
  "provider": {
    "llama.cpp (local)": {
      "models": {
        "qwen2.5-coder-3b": {
          "limit": {
            "context": 16384,
            "output": 14000,
          },
          "name": "Qwen2.5-Coder-3B",
        },
        "qwen2.5-coder-7.6B-Q4_K_M": {
          "name": "Qwen2.5-Coder-7.6B-Q4_K_M",
        },
        "qwen3-ab-3b": {
          "name": "Qwen3-AB-3B",
        },
      },
      "options": {
        "apiKey": "{env:LLAMA_CPP_API_KEY}",
        "baseURL": "http://127.0.0.1:11435/v1",
      },
    },
    "llamastash": {
      "models": {
        "Qwen_Qwen3-4B-Thinking-2507-Q3_K_M": {
          "name": "Qwen_Qwen3-4B-Thinking-2507-Q3_K_M",
        },
      },
      "name": "LlamaStash",
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "apiKey": "{env:LLAMASTASH_API_KEY}",
        "baseURL": "http://127.0.0.1:11435/v1",
      },
    },
    "google-ai": {
      "npm": "@ai-sdk/google",
      "name": "Google AI Studio (free)",
      "options": {
        "apiKey": "{env:GOOGLE_API_KEY}",
      },
      "models": {
        "gemini-3.7-flash": {
          "name": "Gemini 3.7 Flash (free)",
        },
        "gemini-3.5-flash-lite": {
          "name": "Gemini 3.5 Flash Lite (free)",
        },
      },
    },
  },
}