opencode.json

{
  "$schema": "https://opencode.ai/config.json",
  "model": "anthropic/claude-sonnet-4-5-20250929",
  "autoupdate": true,

  "mcp": {
    "exa": {
      "type": "remote",
      "url": "https://mcp.exa.ai/mcp?exaApiKey={env:EXA_API_KEY}&tools=web_search_exa,get_code_context_exa"
    },
    "Ref": {
      "type": "local",
      "command": ["npx", "-y", "ref-tools-mcp@latest"],
      "environment": {
        "REF_API_KEY": "{env:REF_API_KEY}"
      }
    }
  },

  "tools": {
    "Ref*": false,
    "exa*": false
  },

  "permission": {
    "edit": "allow",
    "bash": "allow",
    "webfetch": "allow"
  },

  "agent": {
    "plan": {
      "description": "Strategic planning and coordination agent. Use for: analyzing complex requirements, decomposing tasks into parallelizable work units, architectural decisions, code analysis without modification, creating implementation plans, and coordinating multi-step workflows.",
      "mode": "primary",
      "model": "anthropic/claude-opus-4-5-20251101",
      "temperature": 0.1,
      "prompt": "{file:./.opencode/prompts/plan.md}",
      "tools": {
        "write": false,
        "edit": false,
        "patch": false,
        "bash": false,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": true,
        "Ref*": true,
        "exa*": true
      },
      "permission": {
        "edit": "deny",
        "bash": "deny",
        "webfetch": "allow"
      }
    },

    "build": {
      "description": "Primary implementation agent with full tool access. Use for: writing new code, editing existing files, running build commands, executing tests, installing dependencies, and any task requiring file system modifications or shell access.",
      "mode": "primary",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.2,
      "prompt": "{file:./.opencode/prompts/build.md}",
      "tools": {
        "write": true,
        "edit": true,
        "patch": true,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": true,
        "Ref*": true,
        "exa*": true
      }
    },

    "scout": {
      "description": "Fast codebase exploration agent. Use when you need to: quickly find files by pattern or name, search for specific code patterns or keywords, understand directory structure, locate function or class definitions, or answer questions about what exists in the codebase.",
      "mode": "subagent",
      "model": "anthropic/claude-haiku-4-5-20251001",
      "temperature": 0.0,
      "maxSteps": 15,
      "tools": {
        "write": false,
        "edit": false,
        "patch": false,
        "bash": false,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false
      },
      "permission": {
        "edit": "deny",
        "bash": "deny"
      }
    },

    "research": {
      "description": "Documentation and API research agent. Use when you need to: look up library documentation, find API references, research best practices, understand framework patterns, check package versions or compatibility, or gather information from external sources before implementation.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.3,
      "maxSteps": 25,
      "tools": {
        "write": false,
        "edit": false,
        "patch": false,
        "bash": false,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": true,
        "Ref*": true,
        "exa*": true
      },
      "permission": {
        "edit": "deny",
        "bash": "deny",
        "webfetch": "allow"
      }
    },

    "reviewer": {
      "description": "Code review and quality assessment agent. Use when you need to: review code changes for bugs or issues, assess code quality and maintainability, check for anti-patterns or code smells, evaluate architectural decisions, or provide feedback on pull requests without making changes.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.1,
      "maxSteps": 20,
      "tools": {
        "write": false,
        "edit": false,
        "patch": false,
        "bash": false,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false,
        "Ref*": true
      },
      "permission": {
        "edit": "deny",
        "bash": "deny"
      }
    },

    "tester": {
      "description": "Test creation and validation agent. Use when you need to: write unit tests, integration tests, or end-to-end tests, run existing test suites, analyze test coverage, fix failing tests, or validate that code changes don't break existing functionality.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.2,
      "maxSteps": 25,
      "tools": {
        "write": true,
        "edit": true,
        "patch": false,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false
      },
      "permission": {
        "edit": "allow",
        "bash": {
          "npm test*": "allow",
          "npm run test*": "allow",
          "yarn test*": "allow",
          "pnpm test*": "allow",
          "bun test*": "allow",
          "pytest*": "allow",
          "python -m pytest*": "allow",
          "go test*": "allow",
          "cargo test*": "allow",
          "jest*": "allow",
          "vitest*": "allow",
          "mocha*": "allow",
          "*": "ask"
        }
      }
    },

    "refactor": {
      "description": "Code refactoring agent. Use when you need to: improve code structure without changing behavior, extract methods or classes, rename variables for clarity, reduce code duplication, apply design patterns, or modernize legacy code.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.2,
      "maxSteps": 25,
      "tools": {
        "write": false,
        "edit": true,
        "patch": false,
        "bash": false,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false
      },
      "permission": {
        "edit": "allow",
        "bash": "deny"
      }
    },

    "debug": {
      "description": "Bug investigation and debugging agent. Use when you need to: investigate runtime errors, trace execution flow, analyze stack traces, examine logs, bisect commits to find regressions, or diagnose performance issues.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.2,
      "maxSteps": 25,
      "tools": {
        "write": false,
        "edit": false,
        "patch": false,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false
      },
      "permission": {
        "edit": "deny",
        "bash": {
          "git log*": "allow",
          "git diff*": "allow",
          "git show*": "allow",
          "git blame*": "allow",
          "git bisect*": "allow",
          "cat *": "allow",
          "head *": "allow",
          "tail *": "allow",
          "less *": "allow",
          "grep *": "allow",
          "*": "ask"
        }
      }
    },

    "devops": {
      "description": "Infrastructure and DevOps agent. Use when you need to: write or modify Dockerfiles, create Kubernetes manifests, configure CI/CD pipelines (GitHub Actions, GitLab CI), manage Terraform or infrastructure-as-code, handle cloud provider configurations, set up monitoring or logging, or troubleshoot deployment issues.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.1,
      "maxSteps": 30,
      "tools": {
        "write": true,
        "edit": true,
        "patch": true,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": true,
        "Ref*": true,
        "exa*": true
      },
      "permission": {
        "edit": "allow",
        "bash": {
          "docker build*": "allow",
          "docker compose*": "allow",
          "docker ps*": "allow",
          "docker logs*": "allow",
          "docker images*": "allow",
          "docker inspect*": "allow",
          "kubectl get*": "allow",
          "kubectl describe*": "allow",
          "kubectl logs*": "allow",
          "kubectl apply*": "ask",
          "kubectl delete*": "ask",
          "kubectl exec*": "ask",
          "terraform init*": "allow",
          "terraform plan*": "allow",
          "terraform validate*": "allow",
          "terraform fmt*": "allow",
          "terraform apply*": "ask",
          "terraform destroy*": "deny",
          "aws sts get-caller-identity*": "allow",
          "aws s3 ls*": "allow",
          "aws *": "ask",
          "gcloud config list*": "allow",
          "gcloud *": "ask",
          "az account show*": "allow",
          "az *": "ask",
          "helm list*": "allow",
          "helm status*": "allow",
          "helm *": "ask",
          "*": "ask"
        },
        "webfetch": "allow"
      }
    },

    "security": {
      "description": "Security auditing agent. Use when you need to: scan for vulnerabilities in dependencies, identify security anti-patterns in code, audit authentication or authorization logic, check for secrets or credentials in code, review input validation, assess API security, or generate security reports.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.1,
      "maxSteps": 20,
      "tools": {
        "write": false,
        "edit": false,
        "patch": false,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false,
        "Ref*": true
      },
      "permission": {
        "edit": "deny",
        "bash": {
          "npm audit*": "allow",
          "yarn audit*": "allow",
          "pnpm audit*": "allow",
          "pip-audit*": "allow",
          "safety check*": "allow",
          "trivy *": "allow",
          "grype *": "allow",
          "semgrep *": "allow",
          "snyk *": "allow",
          "bandit *": "allow",
          "gitleaks *": "allow",
          "trufflehog *": "allow",
          "git log*": "allow",
          "git diff*": "allow",
          "*": "deny"
        }
      }
    },

    "docs": {
      "description": "Documentation agent. Use when you need to: write or update README files, create API documentation, generate code comments or docstrings, write architecture decision records, create user guides, or maintain changelog entries.",
      "mode": "subagent",
      "model": "anthropic/claude-haiku-4-5-20251001",
      "temperature": 0.3,
      "maxSteps": 15,
      "tools": {
        "write": true,
        "edit": true,
        "patch": false,
        "bash": false,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false
      },
      "permission": {
        "edit": "allow",
        "bash": "deny"
      }
    },

    "migrate": {
      "description": "Database and schema migration agent. Use when you need to: create or modify database migrations, update ORM schemas (Prisma, Drizzle, TypeORM), write SQL migration scripts, handle data backfills, manage schema versioning, or coordinate database changes across environments.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.1,
      "maxSteps": 25,
      "tools": {
        "write": true,
        "edit": true,
        "patch": false,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false,
        "Ref*": true
      },
      "permission": {
        "edit": "allow",
        "bash": {
          "prisma migrate*": "allow",
          "prisma db*": "allow",
          "prisma generate*": "allow",
          "npx prisma*": "allow",
          "drizzle-kit*": "allow",
          "npx drizzle-kit*": "allow",
          "typeorm migration*": "allow",
          "npx typeorm*": "allow",
          "knex migrate*": "allow",
          "npx knex*": "allow",
          "sequelize db:migrate*": "allow",
          "npx sequelize*": "allow",
          "alembic*": "allow",
          "flask db*": "allow",
          "django-admin migrate*": "allow",
          "python manage.py migrate*": "allow",
          "goose*": "allow",
          "sql-migrate*": "allow",
          "psql*": "ask",
          "mysql*": "ask",
          "sqlite3*": "ask",
          "*": "ask"
        }
      }
    },

    "perf": {
      "description": "Performance profiling and optimization agent. Use when you need to: analyze bundle sizes, run lighthouse audits, profile memory usage, identify performance bottlenecks, optimize database queries, measure load times, or investigate memory leaks.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.2,
      "maxSteps": 25,
      "tools": {
        "write": false,
        "edit": false,
        "patch": false,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false,
        "Ref*": true
      },
      "permission": {
        "edit": "deny",
        "bash": {
          "npm run build*": "allow",
          "npx webpack-bundle-analyzer*": "allow",
          "npx source-map-explorer*": "allow",
          "npx lighthouse*": "allow",
          "lighthouse*": "allow",
          "npx bundlesize*": "allow",
          "npx size-limit*": "allow",
          "node --prof*": "allow",
          "node --inspect*": "allow",
          "clinic*": "allow",
          "npx clinic*": "allow",
          "0x*": "allow",
          "autocannon*": "allow",
          "ab *": "allow",
          "wrk *": "allow",
          "hyperfine*": "allow",
          "py-spy*": "allow",
          "perf *": "allow",
          "time *": "allow",
          "EXPLAIN ANALYZE*": "allow",
          "*": "ask"
        }
      }
    },

    "api": {
      "description": "API design and integration agent. Use when you need to: design REST or GraphQL APIs, create OpenAPI/Swagger specifications, generate API clients or SDKs, implement API versioning, write contract tests, or integrate with external APIs.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.2,
      "maxSteps": 25,
      "tools": {
        "write": true,
        "edit": true,
        "patch": false,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": true,
        "Ref*": true,
        "exa*": true
      },
      "permission": {
        "edit": "allow",
        "bash": {
          "npx @openapitools/openapi-generator-cli*": "allow",
          "openapi-generator*": "allow",
          "npx swagger-cli*": "allow",
          "swagger-codegen*": "allow",
          "npx redocly*": "allow",
          "spectral*": "allow",
          "npx spectral*": "allow",
          "prism*": "allow",
          "npx @stoplight/prism*": "allow",
          "curl*": "allow",
          "httpie*": "allow",
          "http *": "allow",
          "grpcurl*": "allow",
          "graphql-codegen*": "allow",
          "npx graphql-codegen*": "allow",
          "rover*": "allow",
          "*": "ask"
        },
        "webfetch": "allow"
      }
    },

    "data": {
      "description": "Data pipeline and transformation agent. Use when you need to: build ETL pipelines, transform data with pandas/polars, create dbt models, write Spark jobs, design data workflows, or implement data validation and quality checks.",
      "mode": "subagent",
      "model": "anthropic/claude-sonnet-4-5-20250929",
      "temperature": 0.2,
      "maxSteps": 30,
      "tools": {
        "write": true,
        "edit": true,
        "patch": false,
        "bash": true,
        "read": true,
        "grep": true,
        "glob": true,
        "list": true,
        "webfetch": false,
        "Ref*": true
      },
      "permission": {
        "edit": "allow",
        "bash": {
          "dbt run*": "allow",
          "dbt test*": "allow",
          "dbt build*": "allow",
          "dbt compile*": "allow",
          "dbt debug*": "allow",
          "dbt deps*": "allow",
          "python *": "ask",
          "jupyter*": "allow",
          "papermill*": "allow",
          "great_expectations*": "allow",
          "soda*": "allow",
          "dagster*": "ask",
          "prefect*": "ask",
          "airflow*": "ask",
          "spark-submit*": "ask",
          "pyspark*": "ask",
          "*": "ask"
        }
      }
    }
  }
}