一键导入
rp-export-context
Explore the codebase and export a structured context file (e.g., context.xml) for pasting into an external LLM like ChatGPT or Claude Web.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Explore the codebase and export a structured context file (e.g., context.xml) for pasting into an external LLM like ChatGPT or Claude Web.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Gather relevant source files for a task, resolve their dependencies, and package everything into a structured context prompt.
Produce an architectural design and implementation plan based on task requirements, identifying scope, dependencies, impact, and a step-by-step rollout strategy.
Trace execution paths from an error or bug report to find the root cause, gathering surrounding context (callers, imports, tests) along the way.
Analyze dependencies, callers, and coupling before refactoring, then produce a safe incremental refactoring plan.
Review recent code changes by gathering git diffs, identifying affected callers and tests, and analyzing for security, performance, and breaking changes.
Identify untested functions by comparing source symbols with test symbols, then prepare context for writing missing tests.
| name | rp_export_context |
| description | Explore the codebase and export a structured context file (e.g., context.xml) for pasting into an external LLM like ChatGPT or Claude Web. |
Gather relevant project files and package them into a file on disk
(e.g., context.xml) that the user can copy-paste into an external
LLM for planning, review, or implementation.
A file written to the workspace (via build_prompt with output_file)
containing structured project context. The agent reports the file path,
token count, and included files, then STOPS.
explain_architecture / batch_codemap — understand which files are relevantget_imports_graph / blast_radius — trace dependencies to ensure complete contextestimate_tokens — verify context fits the target LLM's context windowbuild_prompt — package and write to file (use output_file="context.xml" or similar)output_file parameter. Never print the full prompt inline.estimate_tokens before build_prompt to avoid exceeding the external LLM's context window.auto_expand_dependencies=True to include imported files automatically.