Enterprise-ai-scenario-map-skill documents a clone command for a different repository
咨询AI Agent Skill - 为任何企业自动生成 AI 应用场景地图报告 | Auto-generate AI scenario map reports for any enterprise
At a glance
- What is it?
- A consulting skill that turns a company name into a 15 to 30 page AI scenario map, with a Python script that emits a list of queries rather than doing research. The setup instructions point somewhere else, and the page arithmetic does not close.
- Who is it for?
- This skill is worth reading for the report structure and the priority framework even if you never run it, because the references are reusable on their own. Before you run anything, resolve two things yourself: the clone command in the documentation does not match this repository, so work out which URL is the one you actually want, and the Python script only produces a list of queries, so the research quality is bounded by the agent you load the skill into.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Activity is slowing. The repository last received commits 6 months ago.
- What is it written in?
- Mainly Python, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on October 4, 2026, and from our analysis. They are not legal advice.
Editorial analysis
The documented clone command clones YuanziAI, not MetaInFLow
The quick start gives three lines, and all three name something other than this repository:
git clone https://github.com/YuanziAI/enterprise-ai-scenario-map.git
cd enterprise-ai-scenario-mapThe repository these instructions live in is `MetaInFLow/Enterprise-ai-scenario-map-skill`. The clone URL points at the `YuanziAI` organisation and at a repository called `enterprise-ai-scenario-map`, without the `-skill` suffix. The directory you land in is likewise `enterprise-ai-scenario-map`, and the same name appears as the root folder in the project structure listing.
So there are two owners and two repository names in play, and a reader following the setup instructions step by step would end up somewhere that is not where they started. Nothing in the document says whether `YuanziAI/enterprise-ai-scenario-map` is a mirror, an older location, or a different project entirely.
The contact block gives a third naming variant. It lists the company as Yuanzi AI, in Shenzhen, with a named contact and a WeChat account, while the repository itself sits under `MetaInFLow`. Three identifiers for one project is the kind of thing worth pinning down before you file an issue against the wrong place.
The part table adds up to 17 pages, while Deep Plan promises 20 to 30
The report is broken into a cover and six parts, each with a stated page count: cover 1, executive summary 1, company profile 3 to 4, industry scan 2, scenario map 5 to 6, implementation path 2, next steps 1.
Add the low ends and you get 15 pages. Add the high ends and you get 17. That is the whole report, and it is the number the structure describes.
The three modes promise something different. Quick Scan is 5 to 8 pages, Standard is 15 to 20 pages, and Deep Plan is 20 to 30 pages. Only Quick Scan and the bottom of the Standard range fit inside the part table. Deep Plan, which is the mode with ROI detail in its description, needs at least three more pages than the structure accounts for and up to thirteen.
So the two tables disagree. The part breakdown is the more specific one, since it allocates page counts per section, while the mode table is a summary. A reader expecting a 30 page document from Deep Plan will find six parts whose stated sizes cannot produce it.
The scenario counts have the same flavour. Part 4 promises a full list of more than 30 AI application scenarios inside 5 to 6 pages, which works out to well under a page per scenario once the priority matrix and three deep-dive scenarios are subtracted.
Three modes scale the scenario count from 15 to 50+, against a library of 64 entries
Mode selection changes the scope. Quick Scan is for finding out what AI could do and yields 15 to 20 scenarios. Standard is for a complete AI adoption plan and yields more than 30. Deep Plan is for a detailed rollout with ROI and yields more than 50.
The built-in reference library underneath is `references/typical-ai-scenarios.md`, described as a library of typical AI scenarios across eight industries. The table lists construction, e-commerce, finance, manufacturing, healthcare, education, legal and logistics, each with 8 or more scenarios. Eight industries at 8 or more is 64 entries in the reference file before anything is generated.
That number does not map cleanly onto the mode table. A 50-plus scenario Deep Plan draws from a reference pool that already holds 64 candidates, so more than half the library is in play at the deepest mode. A 15 to 20 scenario Quick Scan samples under a third of it.
The document is also explicit about what happens for an industry that is not in the table: the agent researches it in real time through web-search and generates scenarios dynamically. So the eight-industry list is a starting library rather than the boundary of what the skill can produce.
The Python script emits a list of queries and the agent does the research
The design decision worth knowing before you install anything is that the script does not research. `scripts/deep_research_wrapper.py` generates a research framework and a list of queries to run. The actual searching is done by the AI agent you loaded the skill into, using its web-search tool.
Three consequences are stated. There are zero external API dependencies and zero additional cost, because the wrapper calls nothing itself. Research quality depends on the capability of the agent you are using. And the skill works on any agent platform that has web-search.
That also explains the prerequisites. Python 3.8 or newer is listed for running the research framework generator script, and the agent with web-search is listed as the other requirement. Python is not doing the analysis; it is producing the questions that the agent then answers.
Two named agents are given as examples: Claude Code, and ChatGPT with browsing. Neither is a dependency. The skill is a markdown file the agent reads, so the platform is whatever you already have.
SKILL.md is the whole product, and the loading step is left undefined
Step two of the usage instructions is to load `SKILL.md` into your AI agent, with the parenthetical note that the method varies by platform. That is the entire install step for the skill itself.
The repository is small enough to inventory. The top level is `.gitignore`, `LICENSE`, `README.md`, `SKILL.md`, a `references/` directory and a `scripts/` directory. Inside `references/` sit six files: the V2.1 report template, a business analysis framework, the eight-industry scenario library, a scenario priority framework, an industry case template and a company-info configuration file. Inside `scripts/` there is one Python file.
So the deliverable is a definition file, six reference documents and one helper script. There is no package to install, no manifest, no build step and no lockfile, which is consistent with the skill being read as text.
The consequence for anyone automating a setup is that step two has no reproducible answer. Different agents load a skill in different ways, and this document does not say which directory it goes in, what metadata the file carries, or how the agent is told to read it. That is fine for a person setting up one agent by hand and unhelpful for a scripted rollout.
MIT with a star request, a V2.1 template and no tagged release
The licence is MIT and a `LICENSE` file sits at the top level, so the reuse terms are the standard permissive ones. The licence section adds that you may use it freely and asks for a star if it helped. The star request is a request, not a condition, and MIT does not require one.
The template is versioned inside the repository rather than by release. `references/report-template-v2.1.md` is the only version marker visible anywhere, which means there is no way to tell from the repository whether v2.1 is current or whether a v3 exists somewhere else. The repository has no GitHub releases, so there is no tag to compare against and no changelog.
Zero issues are open against the repository. That is an unusual figure for a project with 632 stars and a stated willingness to take pull requests, and it cuts both ways: either the questions land elsewhere, or the issue tracker is not where the conversation happens.
The contribution guidance names four things it especially wants: new industry scenario libraries added to `references/typical-ai-scenarios.md`, report template improvements, new language support, and shared example reports. The first is the only one with a named file, which makes it the natural place to start.
Nothing has been pushed since 2026-04-01
The last push to the default branch is dated 2026-04-01. For a repository whose output is a markdown skill that agents read directly, that date matters more than it would for compiled software, because there is no package registry pulling an update and no build to rerun. The version of the skill you have is the version on disk.
Nothing in the document claims a development pace, and none should be inferred. There is no release history, no roadmap file and no changelog to corroborate one.
What the content itself says is useful context rather than a freshness claim. The document positions itself against a stated situation where 90 percent of business leaders have attended a hundred talks and still cannot say what AI would do for them, and against the fear of investing in the wrong direction or missing a window. The scenario map is pitched as a board-meeting artefact rather than a slide deck, and the report structure is built around a priority matrix and a three-stage rollout plan rather than a feature list.
Whether that framing still matches how an enterprise buys AI is a question the artefact cannot answer for you, and the age of the file is worth weighing when you decide how much weight to give it.
Editorial conclusion
This skill is worth reading for the report structure and the priority framework even if you never run it, because the references are reusable on their own. Before you run anything, resolve two things yourself: the clone command in the documentation does not match this repository, so work out which URL is the one you actually want, and the Python script only produces a list of queries, so the research quality is bounded by the agent you load the skill into. And check the date before planning around it, since nothing has been pushed since 2026-04-01.
Frequently asked questions
What is AI mapping and how does it work?
In this skill, mapping means turning a company into a list of AI application scenarios with priorities. The agent researches the company through web-search, applies the business analysis and priority frameworks in the references folder, and writes the result into the V2.1 report template as a scenario map with a priority matrix.
What do I need to run the Enterprise-ai-scenario-map skill?
An AI agent that has a web-search tool, and Python 3.8 or newer for the research framework generator script. The script only produces that list, and the agent does the research, so the agent's capability is what bounds the output quality.
How many pages does the AI scenario map report run to?
The section table gives a cover plus six parts that add up to between 15 and 17 pages. The mode table says Quick Scan is 5 to 8 pages, Standard is 15 to 20, and Deep Plan is 20 to 30, so the Deep Plan range runs past what the section breakdown accounts for.
Which industries does the Enterprise-ai-scenario-map skill cover?
Eight are in the built-in library, each with 8 or more typical scenarios: construction, e-commerce, finance, manufacturing, healthcare, education, legal and logistics. For an industry not on that list, the agent researches it through web-search at run time and generates the scenarios itself.
What license is the Enterprise-ai-scenario-map skill released under?
MIT, with a LICENSE file at the top level of the repository. The README also asks for a star if the skill helps you, which is a request rather than a licence condition.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/metainflow-enterprise-ai-scenario-map-skill)