{"id":"michaljach-opencode-browser","name":"opencode-browser","homepage":null,"repo_url":"https://github.com/michaljach/opencode-browser","category":"automation","subcategories":[],"tags":["browser-automation","web-scraping","web-testing","mcp","opencode","typescript","plugin"],"what_it_does":"Provides an OpenCode plugin that integrates the Browser MCP (browsermcp) to let an AI agent drive a Chrome/Edge browser for automation tasks such as navigation, clicking, form filling, scraping, and screenshot/snapshot-based verification—primarily via MCP tool calls.","use_cases":["Automated browser-based testing flows in an OpenCode agent","Web scraping of page content and extraction of structured data","Automated form filling and submission (e.g., contact/login flows in test environments)","End-to-end UI checks with screenshots/snapshots for verification","Assisted browser navigation from natural-language prompts within OpenCode"],"not_for":["Direct production use for sensitive accounts without additional security review","Fully offline operation where a browser extension and MCP runtime cannot be started","High-integrity automation requiring strict idempotency and transactional safety","Use cases that require a public REST/HTTP API contract independent of OpenCode"],"best_when":"When you already use OpenCode and want an agent to control a real browser via Browser MCP (Chrome/Edge) with guidance that reduces unnecessary tool calls.","avoid_when":"When you need fine-grained audit/logging, strict access control for specific pages/actions, or a documented, standalone API for external systems beyond OpenCode.","alternatives":["Playwright (direct scripting or agent-driven via your own integration)","Selenium WebDriver (scripted automation)","Browser automation frameworks with OpenAI function/tool calling","Standalone MCP client implementations for browser automation (if available) rather than an OpenCode plugin wrapper"],"af_score":63.1,"security_score":39.8,"reliability_score":37.5,"package_type":"mcp_server","discovery_source":["github"],"priority":"high","status":"evaluated","version_evaluated":null,"last_evaluated":"2026-03-30T15:33:46.239525+00:00","interface":{"has_rest_api":false,"has_graphql":false,"has_grpc":false,"has_mcp_server":true,"mcp_server_url":null,"has_sdk":false,"sdk_languages":[],"openapi_spec_url":null,"webhooks":false},"auth":{"methods":[],"oauth":false,"scopes":false,"notes":"No authentication mechanism is described for this plugin itself; it relies on the local Browser MCP server/extension and OpenCode configuration. Any authentication/authorization model would be handled by OpenCode and/or the Browser MCP extension/server."},"pricing":{"model":null,"free_tier_exists":false,"free_tier_limits":null,"paid_tiers":[],"requires_credit_card":false,"estimated_workload_costs":null,"notes":"README does not describe pricing/tiers. Browser automation itself may incur compute/agent usage costs in OpenCode and MCP runtime costs locally."},"requirements":{"requires_signup":false,"requires_credit_card":false,"domain_verification":false,"data_residency":[],"compliance":[],"min_contract":null},"agent_readiness":{"af_score":63.1,"security_score":39.8,"reliability_score":37.5,"mcp_server_quality":55.0,"documentation_accuracy":70.0,"error_message_quality":null,"error_message_notes":"Examples provided for 'connection unavailable' and 'connection restored' plus troubleshooting checklists (extension status, browser running, restart OpenCode). Coverage appears focused on connection issues rather than all failure modes.","auth_complexity":90.0,"rate_limit_clarity":10.0,"tls_enforcement":60.0,"auth_strength":35.0,"scope_granularity":10.0,"dependency_hygiene":40.0,"secret_handling":55.0,"security_notes":"No explicit security controls are documented at the plugin level (no scopes, no per-domain allowlists, no credential handling). The approach controls a real browser via a local MCP server/extension, which increases risk if used with sensitive sites without additional guardrails. README mentions an environment variable (BROWSER_MCP_DEBUG) but does not discuss secret storage/redaction practices. Dependency data is not provided in the manifest, so dependency hygiene cannot be fully verified.","uptime_documented":0.0,"version_stability":55.0,"breaking_changes_history":30.0,"error_recovery":65.0,"idempotency_support":"false","idempotency_notes":"No explicit idempotency guarantees for browser actions (e.g., repeated form submissions or repeated navigation side effects).","pagination_style":"none","retry_guidance_documented":true,"known_agent_gotchas":["Browser state can change between attempts; ensure selectors/actions are specific to avoid unintended interactions","Snapshots/screenshots/waits are described as slower; agents should minimize these unless needed","Connection issues are addressed with immediate retry guidance, but success depends on extension/browser readiness"]}}