Litho, deepwiki-rs and a successor called Terrain: four names for one tool
Turn code into clarity. Generate accurate technical docs and AI-ready context in minutes—perfectly structured for human teams and intelligent agents.
At a glance
- What is it?
- This is a documentation generator written in Rust that reads a codebase and emits architecture diagrams in the C4 model, with the release version, the crate name and the product name all different from each other. The page also announces that the tool has been superseded by a separate project and keeps the old one going as the narrow version. What the manifest adds is the clearest picture of the mechanism: one abstraction crate for model access, and a dependency list annotated in Chinese.
- Who is it for?
- Litho suits a team that wants architecture diagrams regenerated from source rather than maintained by hand, and the C4 framing gives the output a shape a reviewer can check. Three things to weigh before adopting it.
- 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?
- Yes. The repository received new commits within the last day.
- What is it written in?
- Mainly Rust, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- sopaco/deepwiki-rs3,174 ★ · +1,213
| Repository | 2026-09-10 | 2026-11-15 |
|---|---|---|
| sopaco/deepwiki-rs | 2k | 3.2k |
1,213 Stars in the last 66 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
Four names for one tool, and a fifth for its successor
The naming is where this repository starts, and it is worth taking slowly. The repository is deepwiki-rs. The crate published from it is also deepwiki-rs. The product name on the page is Litho, with the heading pairing the two. And the page then announces that Litho has evolved into something else entirely, a separate repository named Terrain, while keeping Litho as what it calls a fast, focused generator for the C4 documentation format. The stated differences are that the successor keeps a knowledge base in sync with code, broadens language and framework support, and is read by mainstream coding agents through a protocol it calls ACP. That is a successor with a different purpose rather than a faster version of the same one, and the predecessor stays online.
The page is mostly HTML tables, and the manifest is annotated in Chinese
The presentation is worth a note because it shapes what a reader sees first. The top of the page is centred markup with a language switch between English and Chinese, four badge links that are duplicated pairs pointing at the same crates.io page and at two documentation directories, and two badges from a third-party ranking site carrying tracking parameters. The comparison between documentation written by hand and documentation generated by the tool is a two-column table with inline styles rather than a list, which is the opposite of what a terminal-rendered markdown reader handles gracefully. Underneath that, the packaging manifest tells a different story about authorship: the crate description is in English, while every section heading in the dependency list is a Chinese comment, from the model layer through networking, the async runtime, serialisation and configuration parsing to a final heading for miscellaneous tools.
The output is four levels of the C4 model and nothing else
The core capability is narrow and that is the project's strength. It analyses a source tree and produces architecture documentation in the C4 model format: context, container, component, and code-level documentation, with diagrams at each level. Alongside that it extracts comments, structures and relationships from the code, and it has a template system so the output shape can be changed. So the generator is not summarising a repository into prose, it is filling in a specific diagram notation, which means the result is comparable between two projects in a way free-form notes are not. The page claims several outcomes from that, among them saving hundreds of hours on manual documentation and meeting compliance requirements with auditable output; both are the page's own claims and neither comes with a measurement attached.
It reads PDFs, SQL and git history, not only source files
The advanced feature list widens the input side considerably. External knowledge integration mounts documentation that is not in your repository, naming PDF, Markdown and SQL files, as knowledge sources for the analysis. Database documentation generates schema documentation with entity relationship diagrams for projects that have a database. Git history is analysed to track how the architecture evolved over time. And the output cross-references code elements against documentation, with embedded diagrams and examples, so a reader can move between a diagram and the code behind it. The mechanism is visible in the manifest: there is a PDF text extraction crate, a regex crate, a TOML parser, a glob matcher and a directory walker, which is the shape of a tool that reads whatever is pointed at it.
Model access goes through a single abstraction crate
The dependency list has one entry that explains what the tool actually does, and it is not an HTTP client to a particular provider. A rig core crate sits at the top, and the description in the manifest says the tool uses large language models to generate the documentation. Everything else in the list is plumbing around that call: an HTTP client with JSON support, a full-featured async runtime, two error handling crates, serialisation with derive support plus a JSON Schema crate, a derive-based command line parser, a TOML parser, date and time handling with serialisation, the directory walker and glob, regular expressions, a hashing crate, base64, the PDF extractor, an async trait helper, a random number generator, a path differencing crate, a markdown crate, futures, a UUID generator with version four support, and the glob matcher again. One crate for the model, everything else for turning a directory into text.
An example configuration ships at the repository root
The root of the repository is short and one entry is a configuration sample: an example file named after the product, sitting beside the manifest and the lockfile. That matters more than it sounds, because a tool this configurable is only usable if you can see what a configuration looks like, and the alternative is reverse-engineering it from the documentation prose. The tree also holds two documentation languages side by side, an assets directory, a skills directory, and two agent instruction directories for two different tools, plus an instruction file whose extension is written in capitals rather than the lowercase convention the rest of the ecosystem uses. There is also a directory named for the successor project inside the predecessor's tree, which suggests the migration is staged here rather than done.
Ecosystem and examples sections are promised and not shown
The last part of the page announces an ecosystem of related tools and stops in the middle of the sentence describing it. Nothing after that point is present here: no member projects, no documentation links for the promised ecosystem, and no worked example of what the generated output actually looks like on a real repository. What is present instead is a stargazers link under a line thanking people, and a sponsor link on the same pattern as most Rust projects. The release history is three versions in six months, with the newest published the day before the last push to the branch, so the version you would install from a registry is close to the code on the branch, which is unusual and convenient.
Editorial conclusion
Litho suits a team that wants architecture diagrams regenerated from source rather than maintained by hand, and the C4 framing gives the output a shape a reviewer can check. Three things to weigh before adopting it. Which name you are installing, since the repository, the crate and the product all differ and the page links a published crate. What it reads, because the feature list goes beyond source files to mounted documentation, SQL schemas and git history, which widens what you have to trust. And which project you are committing to, since the page says this one has evolved into another and describes the differences rather than a migration path.
Frequently asked questions
What is DeepWiki and what does it do, in sopaco/deepwiki-rs?
The repository describes itself as DeepWiki-like and is the tool named Litho, published as the deepwiki-rs crate. Litho is a documentation generation engine that analyses source code and produces architecture documentation in the C4 model format, with context, container, component and code-level diagrams, plus a template system for the output. The page does not describe any other product called DeepWiki.
How do I install or run deepwiki-rs?
The page links a published crate under the name deepwiki-rs on the Rust package registry and points at separate English and Chinese documentation directories. It ships an example configuration file at the repository root. The text available here prints no install command or run command, so the crate registry link and the documentation directories are the two routes it offers.
Which languages does Litho support?
The page claims support for multiple languages, naming Rust, Python, Java, Go, C# and JavaScript among them, and repeats multi-language support in the feature list. The manifest's dependencies back the reading side: a PDF text extractor, a regular expression crate, a TOML parser, a glob matcher and a directory walker.
What has happened to Litho, the tool in sopaco/deepwiki-rs?
The page says Litho has evolved into Terrain, a separate repository, which keeps a knowledge base in sync with code, broadens language and framework support, and is read by mainstream coding agents through a protocol called ACP. Litho stays available as the fast, focused generator for the C4 documentation format, and this repository still carries a directory named for the successor.
What input can Litho use besides source code?
External knowledge integration mounts documentation that is not in your repository, naming PDF, Markdown and SQL files, as knowledge sources. Database schema documentation with entity relationship diagrams is generated for SQL projects, git history is analysed to track architectural evolution, and the output cross-references code elements against documentation with embedded diagrams and examples.
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/sopaco-deepwiki-rs)Star history chart for your README
[](https://hysenlabs.com/projects/sopaco-deepwiki-rs)Was this useful?