{"id":"yukkit-e2b-mcp-server","name":"yukkit-e2b-mcp-server","homepage":"https://github.com/yukkit/e2b-mcp-server","repo_url":"https://github.com/yukkit/e2b-mcp-server/tree/main/packages/python","category":"ai-ml","subcategories":[],"tags":["mcp","e2b","sandbox","code-execution","python","typescript","claude-desktop","tooling","security","devtools"],"what_it_does":"Provides MCP (Model Context Protocol) server tools that let AI clients run code and manage files inside E2B sandboxes. Includes separate JavaScript/TypeScript and Python implementations, offering sandbox lifecycle management, code/command execution, filesystem operations, and optional exposure of sandbox services via public URLs.","use_cases":["Enable code execution for AI assistants in an isolated sandbox","Running Python code / notebook-style execution against user-provided snippets","Executing shell commands for DevOps/diagnostics tasks in isolation","File read/write/list operations within a sandboxed workspace","Launching and sharing sandboxed services via temporary public URLs","Managing multiple concurrent sandboxes with automatic cleanup and limits"],"not_for":["Use as a general remote code execution endpoint without strict sandboxing policies and input controls","Production use cases where you cannot safely handle arbitrary code/commands or data exfiltration risks","Workflows that require documented HTTP/REST APIs, GraphQL, or gRPC endpoints from this package"],"best_when":"You want an MCP-compatible toolset for integrating an AI assistant (e.g., Claude Desktop/Cline) with isolated E2B execution and filesystem capabilities, with client-side configuration via MCP server command/args/env.","avoid_when":"You need strong guarantees about preventing untrusted code from exfiltrating data beyond the intended sandbox boundaries, or you cannot provide/secure the required E2B API key and apply operational limits.","alternatives":["E2B code interpreter / sandbox APIs directly (without MCP)","Other MCP server implementations for code execution","Custom MCP servers wrapping an execution environment (with your own tooling and security controls)"],"af_score":64.2,"security_score":57.5,"reliability_score":30.0,"package_type":"mcp_server","discovery_source":["pypi"],"priority":"low","status":"evaluated","version_evaluated":null,"last_evaluated":"2026-04-04T21:37:13.937488+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":["E2B_API_KEY via environment variable"],"oauth":false,"scopes":false,"notes":"Authentication is described as an E2B API key provided via the MCP server process environment (E2B_API_KEY). No fine-grained scopes or OAuth flow are documented in the provided README."},"pricing":{"model":null,"free_tier_exists":false,"free_tier_limits":null,"paid_tiers":[],"requires_credit_card":false,"estimated_workload_costs":null,"notes":"Pricing for this package is not provided; underlying E2B usage would typically incur separate costs, but no limits/costs are documented here."},"requirements":{"requires_signup":false,"requires_credit_card":false,"domain_verification":false,"data_residency":[],"compliance":[],"min_contract":null},"agent_readiness":{"af_score":64.2,"security_score":57.5,"reliability_score":30.0,"mcp_server_quality":85.0,"documentation_accuracy":75.0,"error_message_quality":0.0,"error_message_notes":null,"auth_complexity":85.0,"rate_limit_clarity":10.0,"tls_enforcement":85.0,"auth_strength":60.0,"scope_granularity":20.0,"dependency_hygiene":50.0,"secret_handling":70.0,"security_notes":"Uses sandbox isolation via E2B (good security posture compared to direct execution). Requires an E2B API key via environment variable; README advises keeping it secure. However, the provided content does not describe granular scopes, key rotation, or rate-limit/abuse controls, and it does not provide details on how tool inputs are constrained beyond a 'secure' sandbox default.","uptime_documented":0.0,"version_stability":40.0,"breaking_changes_history":30.0,"error_recovery":50.0,"idempotency_support":"false","idempotency_notes":null,"pagination_style":"none","retry_guidance_documented":false,"known_agent_gotchas":["The README does not specify rate limits, retry semantics, or idempotency guarantees for tools like run_command/run_code/write_file; agents should implement their own conservative retry/idempotency handling.","Running arbitrary code/commands can have side effects within the sandbox; agents should explicitly manage sandbox lifecycle (create/kill) to avoid resource leakage.","Two implementations exist (JS and Python); ensure you target the correct client/tooling configuration expected by your MCP host."]}}