{"ok":true,"host":"execbox.xyz","status":"ready","manifest":{"positioning":"ExecBox is a remote code execution host targeting agents that need to compile and run untrusted or dynamically generated code in an isolated environment. It serves use cases such as running algorithm outputs, evaluating code snippets, or benchmarking numeric workloads. It is distinguished by two resource tiers rather than a single endpoint, allowing cost-appropriate selection based on workload intensity.","host_overview":"ExecBox provides remote sandboxed code execution across multiple programming languages, returning structured JSON output including stdout, stderr, exit code, timing, and memory usage. It offers two execution tiers differentiated by CPU timeout (15s vs 30s) and RAM allocation (512MB vs 1GB). Network access is disabled in all sandboxes.","routing_guidance":"Use ExecBox when an agent needs to execute source code remotely in a sandboxed environment and requires structured output (stdout, stderr, exit code, timing, memory). Choose execute-code-compute-tier for lightweight scripts and quick computations; choose execute-heavy-code for recursive algorithms, benchmarks, or memory-intensive workloads. Do not use ExecBox for any code that requires network access — sockets and HTTP requests are blocked in all tiers. Do not use it for long-running jobs exceeding 30 seconds or workloads requiring more than 1GB RAM; for those, a full cloud compute service (e.g. AWS Lambda with custom limits, Google Cloud Run, or a dedicated VM) would be more appropriate. ExecBox does not provide persistent storage, scheduling, or stateful execution between calls.","capability_clusters":[{"skill_names":["execute-code-compute-tier","execute-heavy-code"],"cluster_name":"Sandboxed Code Execution","cluster_summary":"Remotely compiles and runs source code in an isolated environment, returning structured output including stdout, stderr, exit code, wall-clock time, and memory usage. Two tiers are available: a standard tier (15s CPU, 512MB RAM) and a heavy tier (30s CPU, 1GB RAM)."}],"cross_skill_workflows":[{"steps":[{"skill_name":"execute-code-compute-tier","description":"Attempt execution on the standard tier (15s CPU, 512MB RAM) to minimize cost for typical workloads."},{"skill_name":"execute-heavy-code","description":"If the standard tier returns a timeout exit code or memory error, re-submit the same code to the heavy tier (30s CPU, 1GB RAM) to obtain a successful result."}],"when_to_use":"Use when an agent needs to run code but is uncertain whether it will complete within the standard resource limits — attempt the cheaper tier first and escalate if it times out or exceeds memory.","workflow_name":"Tiered Execution with Fallback"}]},"model":"claude-sonnet-4-6","version_no":1,"generated_at":"2026-07-20T18:47:17.809Z","provenance":"ai_authored_unreviewed","ai_authored":true,"merchant_reviewed":false,"merchant_edited":false,"merchant_reviewed_at":null,"merchant_edited_at":null,"skill_md_url":"https://x402gle.com/servers/execbox.xyz/SKILL.md","skills_url":"https://x402gle.com/servers/execbox.xyz/skills.json"}