把自然语言描述转为结构化 JSON,并由 mcp-diagram-generator MCP 服务生成 Draw.io、Mermaid 或 Excalidraw 图表文件。
文档
Accounting Finance System Research Skill
试用Research and solve "how do I do this?" questions inside accounting and finance software systems (ERP, GL, AP/AR, billing, close, and reporting tools). Use wh...
它能做什么
Follow a fixed process for system-how-to support: collect facts, ask clarifying questions, confirm output format, confirm understanding, research external guidance, analyze the best path, and generate a DOCX deliverable.
技能文档
Accounting And Finance System Research
Overview
Follow a fixed process for system-how-to support: collect facts, ask clarifying questions, confirm output format, confirm understanding, research external guidance, analyze the best path, and generate a DOCX deliverable.
Required Behavior
- Ask clarifying questions before proposing a solution.
- Confirm whether output should be
quick memoorsimple q-and-a. - Restate understanding and wait for confirmation before web research.
- Research the internet after confirmation, prioritizing official vendor guidance.
- Separate source-backed guidance from assumptions or inference.
- Include source links and accessed dates in the deliverable.
- Generate a DOCX report in the user-selected format.
Workflow
1) Intake And Scope
- Capture the user objective in one sentence.
- Confirm system name and version/edition (for example: NetSuite, SAP S/4HANA Cloud, Dynamics 365 Finance, QuickBooks).
- Confirm module or workflow area (for example: AP, AR, close, reconciliations, reporting).
- Confirm role/permission constraints and any deadline pressure.
2) Clarification Questions (Mandatory)
- Use references/clarification-question-bank.md.
- Ask only missing critical questions; avoid redundant prompts.
- Pause solutioning until enough facts are available.
- If facts remain unknown, continue with explicit assumptions and conditional guidance.
3) Output Format Confirmation (Mandatory)
- Ask the user to choose one:
quick memo: concise professional summary with recommendation and steps.simple q-and-a: direct answer format with numbered actions.- Default to
quick memoonly when user gives no preference.
4) Understanding Confirmation (Mandatory)
- Restate:
- problem statement
- system context
- open questions or assumptions
- chosen output format
- Ask for explicit confirmation before researching.
5) Research External Guidance
- Follow source ranking in references/source-priority.md.
- Gather at least two relevant sources where possible.
- Include at least one official vendor source when available.
- Track per-source metadata: title, publisher, URL, updated/published date if available, accessed date.
- Prefer current version-specific guidance over older generic content.
6) Analyze And Build Recommendations
- Translate research into concrete user actions.
- Provide prerequisites and permissions needed for each action.
- Add fallback steps when primary path is blocked.
- Include validation checks to confirm completion.
- Call out risks and unresolved dependencies.
7) Generate DOCX Deliverable
- Build JSON input using references/report-json-schema.md.
- Run:
python scripts/build_system_guidance_docx.py \
--input-json \
--output-docx \
--format
- Map
quick memotomemo. - Map
simple q-and-atoq-and-a. - Confirm the document includes:
- request summary and system context
- clarifications and assumptions
- guidance sources with URLs
- recommended steps
- validation checks
- risks and open items
8) Quality Check
- Verify recommendations are consistent with cited guidance.
- Verify assumptions are explicit and easy to review.
- Verify deliverable format matches user request.
- Verify every source has a URL and accessed date.
- Verify output file is
.docxand readable.
Resources
- Clarification checklist: references/clarification-question-bank.md
- Source hierarchy: references/source-priority.md
- Report JSON schema: references/report-json-schema.md
- Example report payload: references/example_report_input.json
- DOCX generator:
scripts/build_system_guidance_docx.py
Dependency
Install once if needed:
python -m pip install --user python-docx
相关技能
在本地磁盘以分类纯 Markdown 文件保存需要长期留存的事实,与智能体内置记忆并存。
诊断生产力系统反复失效的根因,给出最小干预——容量测算、瓶颈定位、可靠的本地记录。
以 AI 机器人身份加入视频会议,提供语音、虚拟形象与屏幕共享四种模式。
从 AdMapix API 拉取广告创意、应用、榜单和收入预估等数据,原样返回结构化 JSON。
按用户明确指令,在得到大脑(Get笔记)中保存、搜索并管理笔记与知识库。
chipmunkrpa 的更多技能
浏览全部技能Research technical accounting treatment and financial statement disclosure for specific transactions using U.S. GAAP and SEC-focused sources. Use when a user...
Review Business Requirements Documents in `.docx` format by reading the existing BRD, extracting paragraph-level context, drafting clarification questions fo...
Review U.S. individual income tax returns (Form 1040/1040-SR) for the most recent tax year, compare major return items against current-year tax rules, check...
Develop deposition question sets from Relativity-exported PDF productions using a user-provided legal theory. Use when tasks involve reviewing opponent-produ...
Review and redline DOCX contracts paragraph by paragraph with tracked changes, clause-level risk analysis, and draft comment responses. Use when a user wants...
Finance, accounting, and investment research automation via the FinResearchClaw repo. Use when asked to run autonomous finance research workflows such as eve...