“code execution and sandboxes”

Create an idempotent PatchDraft only in an isolated PatchFlow workspace.

Delegate bounded support work without replacing the delivery owner or orchestrator.

Use when creating, advancing, or closing a milestone execution plan.

Guide for running Python code on Modal, the serverless compute platform for AI workloads, batch jobs, scheduled tasks, web endpoints, and sandboxed code…

Implement or resume work governed by an existing plan, coordinate explicitly authorized delegated Tasks for named plan files, maintain progress, perform…

执行 Python 代码以进行数学计算、数据处理、字符串操作或算法验证。当遇到复杂的逻辑问题或需要精确计算时使用。

Audit Medis Touch broker capability, risk, portfolio admission, execution ledger and latency/accuracy telemetry.

Use when evaluating any new tool, mechanism or third-party skill, or when tempted to stop and ask mid-run.

Activate for maximal browser and sandbox computer utilization interactive page control file system operations code execution and extended compute.

Plan a non-trivial feature, bug fix, or workflow change in docs/exec-plan/todo.

Fail-closed investment workflow automation for data prewarm, market heartbeat/alerts, monitor state markers, and explicit-confirmation report pushes.

Estimate spread/slippage/fee friction and net expected R before execution.

Guidelines for selecting and running AWX job templates manually.

Vercel Sandbox guidance — ephemeral Firecracker microVMs for running untrusted code safely.

Design, implement, or review workers, test runners, evaluators and plugin hosts that execute less-trusted commands or code.

Expert guidance on implementing and managing background tasks in modern Android applications (API 30+).

Transforms analytical outputs into concise, executive-ready communication from a Finance Business Partner perspective.

DEPRECATED — split into execution-core (boot) + execution-extras (routing/lessons).

Help user prepare docker-compose under workspace compose/ when they ask.

Execute code safely using RLM Runtime with Docker isolation when you need to run tests, validate implementations, or execute complex computations.

A股算法执行/智能拆单策略。当用户说"算法执行"、"智能拆单"、"execution algo"、"算法交易"、"拆单策略"、"大单拆分"时触发。基于 cn-stock-data 获取数据，设计算法执行方案。支持 formal/brief 两种输出风格。

Task implementation workflow, test harnesses, screenshot/video capture, visual debugging.

AutoForge is a production-grade autonomous optimization framework for AI agents.

Pipeline routing, classification, phase ownership, principles, lessons protocol.