> ## Documentation Index > Fetch the complete documentation index at: https://cognis.vasanth.xyz/llms.txt > Use this file to discover all available pages before exploring further. # Ways to contribute > Bugs, features, integrations, docs, examples — every contribution lane. Thanks for considering a contribution. Cognis is a young project; almost every kind of help is welcome. This page lists the most common contribution shapes and where to start with each. ## Report a bug You hit something that doesn't work. Open an issue with: * The Cognis version (`cognis = "0.3.x"` or commit hash if from main). * A minimal reproducing example. The smaller, the faster we can fix it. * What you expected vs what happened. * Rust toolchain version (`rustc --version`). Bug reports go to [GitHub Issues](https://github.com/0xvasanth/cognis/issues). Use the Bug Report template if it's available. ## Request a feature Open an issue describing: * The problem you're trying to solve. * The shape you'd want the API to take. * Why existing primitives don't cover it. For larger ideas, open a [Discussion](https://github.com/0xvasanth/cognis/discussions) first to align on direction before writing code. ## Improve the docs Documentation contributions are first-class. Three flavors: * **Fix typos and rough edges** — open a PR directly. * **Sharpen explanations** — if a page felt confusing, rewrite the part you got stuck on. You're closer to that experience than the maintainers. * **Add missing pages** — propose a topic in an issue first; the docs site has a structure (Learn / Reference / Examples / Contribute) that we try to keep coherent. The docs source lives at [`docs/mintlify/`](https://github.com/0xvasanth/cognis/tree/main/docs/mintlify). It's [Mintlify](https://mintlify.com) — write Markdown / MDX with Cognis-flavored conventions. ## Add a provider Cognis is provider-pluggable by design. To add a new LLM provider (or vector store, embedder, checkpointer): | Adding a… | Read | | ------------ | -------------------------------------------------------------- | | LLM provider | [Adding a new provider](/contribute/adding-a-provider) | | Vector store | [Adding a new vector store](/contribute/adding-a-vector-store) | | Tool | [Adding a new tool](/contribute/adding-a-tool) | Each has a step-by-step. The TL;DR: implement the trait, gate behind a feature flag, add tests, add an example, update docs. ## Contribute code For non-trivial changes, the workflow is: 1. Open an issue or discussion describing the change. 2. Fork, branch, implement. 3. Run the [pre-push checklist](/contribute/development-setup#pre-push-checklist). 4. Open a PR following the [PR guidelines](/contribute/pull-requests). 5. Iterate on review. For tiny fixes (typos, missing imports, broken links), skip the issue and open the PR directly. ## Build an integration Cognis pairs well with deployment platforms, observability tools, and frontend frameworks. If you build something that integrates Cognis (a Cognis-flavored crate, a hosted service connector, an Axum/Actix middleware), tell us — it can land on the docs site under integrations or be highlighted in releases. ## Help triage Cognis maintains a small core team. Triage help is high-leverage: * Reproduce reported bugs and confirm they exist. * Label issues by area (`agent`, `graph`, `rag`, `trace`, `docs`). * Close stale duplicates with a friendly pointer. You don't need write access to do this — adding a comment with the reproduction and labels you'd suggest is genuinely useful. ## Be respectful Be kind. We're building something together. Disagreement is fine; condescension and trolling aren't. See the [Code of conduct](/contribute/code-of-conduct). ## See also Local build, test, and the pre-push checklist. Title format, description checklist, sign-off. Crate boundaries and design rules.