AgentSkillsCN

x64dbg-commands-agent

当您需要实现或映射x64dbg的命令技能与代理，安全封装命令执行过程，或将已文档化的x64dbg命令转化为MCP工具的使用方法与安全的工作流程时，可使用此技能。

SKILL.md
--- frontmatter
name: x64dbg-commands-agent
description: Use when implementing or mapping x64dbg command skills/agents, wrapping command execution safely, or translating documented x64dbg commands into MCP tool usage and safe workflows.

x64dbg Commands Agent

Scope

Use this skill to turn x64dbg command documentation into concrete MCP workflows. Prioritize safety wrappers (pause/confirm/rollback) and prefer explicit VA expressions.

References

  • •Read references/commands.md for command lists, syntax, and wrapper guidance.

Workflow

  1. •Identify the command category and exact command name from the reference.
  2. •Decide if the operation is write/destructive; if yes, require pause + confirmation.
  3. •Prefer CommandRun (Python wrapper) for safe execution with optional snapshots.
  4. •For reads, prefer detailed APIs (e.g., MemoryReadDetailed) when partial reads matter.
  5. •For pause/step: call DebugPause(wait=true, timeoutMs=30000) and use DebugStep* with auto_pause=true to avoid “Debugger running” errors.
  6. •For run‑to‑user: call Debug/RunUntilUserCode(wait=true, timeoutMs=30000, pauseFirst=true) or ExecCommand("RunToUserCode"); re‑check RIP via DisasmGetInstructionAtRIP.

Safety defaults

  • •Pause before writes (register/memory/flags/breakpoints).
  • •Use confirm=true for write endpoints in safe mode (register/memory/flags/cmdline/stack/asm/stop).
  • •Use dry_run or require_confirm when risky.
  • •Capture before/after state if values may need rollback.

Notes

  • •Keep SKILL.md lean; load references/commands.md only when needed.