# mesa-dev
Rust SDK for the [mesa.dev](https://mesa.dev) API — manage repositories, branches, commits, content, diffs, and API keys programmatically.
[API Documentation](https://docs.rs/mesa-dev)
## Installation
```sh
cargo add mesa-dev
```
## Quick Start
```rust,no_run
use mesa_dev::{Mesa, MesaError, models::CreateRepoRequest};
#[tokio::main]
async fn main() -> Result<(), MesaError> {
let client = Mesa::new("my-api-key");
// Create a repository
let repo = client
.repos("my-org")
.create(&CreateRepoRequest {
name: "my-repo".to_owned(),
default_branch: None,
})
.await?;
println!("Created repo: {}", repo.name);
// List all branches (automatically paginates)
let branches = client
.branches("my-org", "my-repo")
.list_all()
.collect()
.await?;
println!("Found {} branches", branches.len());
Ok(())
}
```
## Resources
| Repos | `client.repos(org)` | create, list, get, rename, delete |
| Branches | `client.branches(org, repo)` | create, list, delete |
| Commits | `client.commits(org, repo)` | create, list, get |
| Content | `client.content(org, repo)` | get (files and directories) |
| Diffs | `client.diffs(org, repo)` | get |
| Admin | `client.admin(org)` | API key management |
Paginated endpoints expose a `list_all()` method that returns a `PageStream` with lazy cursor-based iteration.
## HTTP Backends
The SDK is generic over its HTTP transport via the `HttpClient` trait.
| `reqwest-client` | [reqwest](https://docs.rs/reqwest) | yes | yes |
| `ureq-client` | [ureq](https://docs.rs/ureq) | no | no |
You can also bring your own backend by implementing `HttpClient`. See the [trait documentation](https://docs.rs/mesa-dev/latest/mesa_dev/trait.HttpClient.html) for a full guide.
## Retry
All requests are retried with exponential backoff and jitter (up to 3 attempts by default). Retryable conditions: HTTP 429, 5xx responses, timeouts, and connection errors. Retry parameters are configurable via `ClientBuilder`.
## Minimum Supported Rust Version
Rust **1.85+** (edition 2024).
---
> **Note:** A large part of this documentation was generated with the assistance of an LLM. If you spot inaccuracies, please open an issue.