yalongwastaken/git-koi

What it does

Git koi generates pixel-art koi from a GitHub account's public activity. Language data colors the fish, contributions determine its growth, streaks mark it, and the previous day's activity decides how it swims. Each run produces a looping GIF suitable for a profile README.

The same generator ships in three forms: a GitHub Action for profile repositories, a git-koi command-line tool, and the Rust library behind both, published on crates.io. The README shows six bodies (white, black, Asagi, tea, gold, platinum) at each growth stage.

Generate a koi

The getting-started guide covers all three paths.

Profile README with the Action: add a workflow in your profile repository that uses the bundled action and its built-in GITHUB_TOKEN, then reference the generated GIF from your README. The author's own profile workflow and GIF are in the example profile workflow and the rendered koi.

Command line: get a binary from the releases page, or build from source with Rust at or above the rust-version in Cargo.toml. The CLI requires the GitHub CLI, authenticated, to read account activity. Run the same command locally to produce a GIF outside GitHub Actions.

Library: depend on the crate and call its API; the reference is on docs.rs.

Constraints

• Profile-README rendering assumes a GitHub profile repository whose workflow can use its built-in token; the Action is tied to that environment.
• Building from source requires Rust at or above the version pinned in Cargo.toml, and the CLI additionally needs an authenticated GitHub CLI session.
• The output is a GIF, so the fish is a 2D pixel-art animation rather than a static image or interactive widget.
• Released under the MIT license.

HotRepo's take

On 2026-11-01 the repository appeared on the Rust new list with 111 stars gained in an approximately eight-hour window starting 01:45 UTC, reaching 213 stars by 09:45 UTC; the hourly samples show a steady climb over that morning rather than a single spike. No all-language daily rank was recorded, and no discussion threads were provided. The project fits the current appetite for generative profile decorations, and shipping the same renderer as a CLI and a crate lets people try it without committing to the Action first.